oracle® jdbc for rdbdownload.oracle.com/otndocs/products/rdb/pdf/rdbjdbc_7341_rn.pdf · 4.2 using...

Oracle® JDBC for Rdb Release Notes Release 7.3.4.0.5 February 2016

Oracle JDBC for Rdb Release Notes, Release 7.3.4.1.0

Copyright © 2005, 2016 Oracle and/or its affiliates. All rights reserved.

Primary Author: Jim Murray.

Contributing Author:

Contributor: Wolfgang Kobarg-Sachsse.

This software and related documentation are provided under a license agreement

containing restrictions on use and disclosure and are protected by intellectual property

laws. Except as expressly permitted in your license agreement or allowed by law, you

may not use, copy, reproduce, translate, broadcast, modify, license, transmit, distribute,

exhibit, perform, publish, or display any part, in any form, or by any means. Reverse

engineering, disassembly, or decompilation of this software, unless required by law for

interoperability, is prohibited.

The information contained herein is subject to change without notice and is not warranted

to be error-free. If you find any errors, please report them to us in writing.

If this is software or related documentation that is delivered to the U.S. Government or

anyone licensing it on behalf of the U.S. Government, the following notice is applicable:

U.S. GOVERNMENT RIGHTS Programs, software, databases, and related

documentation and technical data delivered to U.S. Government customers are

"commercial computer software" or "commercial technical data" pursuant to the

applicable Federal Acquisition Regulation and agency-specific supplemental regulations.

As such, the use, duplication, disclosure, modification, and adaptation shall be subject to

the restrictions and license terms set forth in the applicable Government contract, and, to

the extent applicable by the terms of the Government contract, the additional rights set

forth in FAR 52.227-19, Commercial Computer Software License (December 2007).

Oracle America, Inc., 500 Oracle Parkway, Redwood City, CA 94065.

This software or hardware is developed for general use in a variety of information

management applications. It is not developed or intended for use in any inherently

dangerous applications, including applications that may create a risk of personal injury. If

you use this software or hardware in dangerous applications, then you shall be

responsible to take all appropriate fail-safe, backup, redundancy, and other measures to

2

ensure its safe use. Oracle Corporation and its affiliates disclaim any liability for any

damages caused by use of this software or hardware in dangerous applications.

Oracle and Java are registered trademarks of Oracle and/or its affiliates. Other names

may be trademarks of their respective owners.

Intel and Intel Xeon are trademarks or registered trademarks of Intel Corporation. All

SPARC trademarks are used under license and are trademarks or registered trademarks of

SPARC International, Inc. AMD, Opteron, the AMD logo, and the AMD Opteron logo

are trademarks or registered trademarks of Advanced Micro Devices. UNIX is a

registered trademark of The Open Group.

This software or hardware and documentation may provide access to or information on

content, products, and services from third parties. Oracle Corporation and its affiliates are

not responsible for and expressly disclaim all warranties of any kind with respect to third-

party content, products, and services. Oracle Corporation and its affiliates will not be

responsible for any loss, costs, or damages incurred due to your access to or use of third-

party content, products, or services.

3

Contents Preface................................................................................................................................. 9

Purpose of This Manual .................................................................................................. 9 Intended Audience .......................................................................................................... 9

Access to Oracle Support ................................................................................................ 9 Document Structure ........................................................................................................ 9 Conventions .................................................................................................................... 9

Chapter 1 Installation and Documentation .................................................................... 11

1.1 Accessing the Documentation................................................................................. 11 1.2 System and Software Requirements ....................................................................... 12 1.3 Installation............................................................................................................... 13

1.3.1 Remove Prior BETA Versions of Oracle JDBC for Rdb Release 7.3.x.x.x ... 13

1.3.2 Contents of the Oracle JDBC for Rdb Kit ....................................................... 14 1.3.3 Installation Procedure ...................................................................................... 16 1.3.4 Rdb Thin driver software on Windows ............................................................ 20

1.3.5 Rdb Thin driver software on Other operating systems .................................... 21 Chapter 2 Enhancements Provided in Oracle JDBC for Rdb Release 7.3.4.1.x. .......... 22

2.1 New Transaction Option AUTOFETCH ................................................................ 22

Chapter 3 Problems Corrected ....................................................................................... 24

3.1 Pool Servers Not Working ...................................................................................... 24 3.2 Problem using Multiprocess Connection Option with the Native JDBC driver ..... 24 3.3 Bumpy-case Connection String Options used with Properties. .............................. 24

3.4 Further Blob problems with Delimited Identifiers.................................................. 25 3.5 Access Violation at RDBJDBCMPSHR73+00227E31 .......................................... 26

Chapter 4 Known Problems and Workarounds ............................................................. 28 4.1 Thin Server Deadlocks ............................................................................................ 28 4.2 Using Java Fast VM on OpenVMS ALPHA .......................................................... 28

4.3 Using the Oracle SQL/Services Management GUI and JDBC Dispatchers ........... 29 4.4 Blob Columns and Correlation Names ................................................................... 29

4.5 Blob Columns and Update Statements ................................................................... 30 4.6 External Procedures and Thin Server ..................................................................... 31

4.7 Limitations .............................................................................................................. 31 4.7.1 Unsupported Methods ...................................................................................... 31 4.7.2 Auto-generated keys ........................................................................................ 33 4.7.3 String Truncation Warnings ............................................................................. 33 4.7.4 Numeric and String Functions in JDBC .......................................................... 33

Chapter 5 New Features and Corrections in Previous Releases .................................... 34 5.1 New Features for Release 7.3.4.0.0 ........................................................................ 34

5.1.1 Create / Drop Database Support ...................................................................... 34

5.2 Corrections in Release 7.3.4.0.0 ............................................................................. 34 5.2.1 Like Escape Throws OutOfBounds Exception ................................................ 35 5.2.2 Correlation Name Workaround for Blobs Fails ............................................... 35 5.2.3 PreparedStatement Cache Failure .................................................................... 35

5.2.4 Blob problem with Delimited Identifiers ......................................................... 36 5.3 New Features for Release 7.3.3.2.x ........................................................................ 36

4

5.3.1 Change Transaction Start Time. ...................................................................... 36 5.4 Corrections in Release 7.3.3.2.0 ............................................................................. 36

5.4.1 NullPointerException when allowing Anonymous ......................................... 37 5.4.2 UnsupportedEncodingException: Binary ........................................................ 37

5.4.3 Alignment Faults when using MP Server ........................................................ 37 5.4.4 ResultSet.absolute() Problems ......................................................................... 38 5.4.5 Orphaned Executor Processes when using Persona ......................................... 38

5.5 New Features for Release 7.3.3.1.0 ........................................................................ 39 5.5.1 Change in Template Configuration Filename .................................................. 39

5.6 Corrections in Release 7.3.3.1.0 ............................................................................. 40

5.6.1 DatabaseMetadata.getColumns Problem with DefaultValue .......................... 40

5.6.2 Show Server Configuration file Problem ......................................................... 40 5.6.3 Unknown Action Message in Log file ............................................................. 41 5.6.4 Severe Performance Degradation in MP Server .............................................. 41 5.6.5 Controller DCL Command line Failure ........................................................... 41

5.6.6 Server Access Restrictions Ignored ................................................................. 42 5.6.7 Datetime Insertion Issues ................................................................................ 43

5.6.8 MP Server may Become CPU Bound .............................................................. 44 5.6.9 Controller Null Pointer Exception when no Default Server ............................ 44 5.6.10 Pool Server Tries to Restart Running Server ................................................. 45

5.6.11 Controller Poll Reopenlogs Hangs................................................................. 46

5.6.12 Pool Server Usage Balancing Problem .......................................................... 47 5.6.13 Configuration File Problem Regression......................................................... 47 5.6.14 Server Configuration Deny User not Enforced .............................................. 47

5.7 New Features for Release 7.3.3.0.x. ....................................................................... 48 5.7.1 Oracle Rdb PID now Displayed ....................................................................... 48

5.7.2 Controller Command Show Executors ............................................................ 49 5.7.3 Extra Timestamp Precision .............................................................................. 49 5.7.4 Nlslang Connection Switch.............................................................................. 50

5.7.5 Use Query Header as Description .................................................................... 51 5.8 Corrections in Release 7.3.3.0.x ............................................................................. 52

5.8.1 Controller START SERVER Problems ........................................................... 52 5.8.2 Executor Initialization Problem ....................................................................... 53 5.8.3 Idle Client Termination breaks Subsequent Connection ................................. 53 5.8.4 Controller STOPSERVER command Problem ................................................ 54

5.8.5 Hang on Connection to Server when using SSL .............................................. 54 5.8.6 DatabaseMetadata Pattern Matches and Nulls ................................................. 55 5.8.7 DatabaseMetadata Missing Index Information ................................................ 55 5.8.8 Forced Client Termination may Crash Thin Server ......................................... 56 5.8.9 Show Clients may cause NullPointerException in Server ............................... 56

5.8.10 AccessViolation in MP Server ....................................................................... 57

5.8.11 Syntax Errors in Insert and Update Statements ............................................. 58

5.8.12 Statement.getGeneratedKeys() throws RdbException ................................... 58 5.8.13 SQLException thrown in ResultSet.isBeforeFirst() ...................................... 59 5.8.14 PreparedStatement executeBatch() Problems ................................................ 59

5.9 New Features for Release 7.3.2.0.x ........................................................................ 60

5

5.9.1 Server Options List Inheritance ....................................................................... 60 5.9.2 Oracle JDBC for Rdb Manager Server ............................................................ 61 5.9.3 Restricting Server Access by IP ....................................................................... 62 5.9.4 Executor Balancing .......................................................................................... 62

5.9.5 MinFreeExecutors ............................................................................................ 62 5.9.6 Executor Reuse ................................................................................................ 63 5.9.7 LogFile Patterns ............................................................................................... 63 5.9.8 New Server Configuration Option retainRdbSQLState ................................... 63 5.9.9 New Connection Option app ........................................................................... 64

5.10 Corrections in Release 7.3.2.0.x ........................................................................... 64

5.10.1 Event Flag Problem with long Executor name Prefixes ................................ 64

5.10.2 Executor Process Termination Problem......................................................... 65 5.10.3 Some Server Characteristics not Correctly Inherited from DEFAULT......... 66 5.10.4 Memory Problem with Pool Servers and Java 1.6.0-2................................... 67 5.10.5 Exception not Thrown when Record Locked during Update ........................ 68

5.10.6 Classpath Documentation Error ..................................................................... 68 5.11 New Features for Release 7.3.1.0.x ...................................................................... 69

5.11.1 “Owner” may be used in XML Configuration Files ...................................... 69 5.11.2 SQL Statement Restriction and Denial .......................................................... 70 5.11.3 Event Notification .......................................................................................... 70

5.12 Corrections in Release 7.3.1.0.x ........................................................................... 70

5.12.1 Multi-process Server Connection Hang when Executor Dies ....................... 71 5.12.2 Multi-process Executor Process Terminates Unexpectedly .......................... 71 5.12.3 Pooled Server AutoRestart Problem .............................................................. 72

5.12.4 getGeneratedKeys() and OutOfBounds Exception ........................................ 72 5.12.5 Multi-process Server fails when Persona Used ............................................. 73

5.12.6 RestrictAccess being Applied even when Disabled ....................................... 74 5.12.7 Multi-process Problem in the Native Driver .................................................. 75

5.13 New Features for Release 7.3.0.2.x ...................................................................... 75

5.13.1 Server Network Keep Alive ........................................................................... 75 5.14 Corrections in Release 7.3.0.2.x ........................................................................... 76

5.14.1 Null pointer Exception during Server viability check by Pool Server........... 76 5.14.2 MINUS and INTERSECT Problem ............................................................... 76 5.14.3 Memory Leak when Executor Process fails to Run ....................................... 76 5.14.4 Lost Executor not caught by Multi-process Server........................................ 77

5.14.5 Multi-process Native Driver Problem ............................................................ 77 5.14.6 PreparedStatement.getGeneratedKeys() Problem.......................................... 77 5.14.7 Excessive IOs during Metadata Retrieval ...................................................... 78 5.14.8 Syntax error SQL-F-CONVARUND ............................................................. 78 5.14.9 Multi-process Server Hang ............................................................................ 78

5.14.10 SSL Socket Intrusion Problem ..................................................................... 79

5.15 New Features for Release 7.3.0.1.x ...................................................................... 79

5.15.1 Reopen Server Log files using Poll Subcommand ........................................ 79 5.16 Corrections in Release 7.3.0.1.x ........................................................................... 80

5.16.1 Show Clients not Showing Column Names ................................................... 80 5.16.2 Batched Statement Fails with MULTIPLE_RECORDS Exception .............. 81

6

5.16.3 Incompatibility between 7.3.0.0 Thin Driver and Prior Release Servers ...... 82 5.16.4 Global Memory leak when Executors are Run-down .................................... 82 5.16.5 Autorestart on Pooled Servers not Working .................................................. 83 5.16.6 setFetchSize() hint Ignored by PreparedStatement ........................................ 83

5.16.7 Missing Stmt Exception on Subsequent Execution of PreparedStatement .... 84 5.16.8 EOFException on READ_ROW with Nested Statements ............................. 84 5.16.9 DatabaseMetaData.getUDTs() ....................................................................... 85 5.16.10 Using Multi-process with the JDBC Native Driver ..................................... 86 5.16.11 Prepared Statement not Closing underlying Cursor ..................................... 86

5.16.12 Release Statement Synchronization Problem .............................................. 87

5.16.13 Unitialized SQLCA Block in MP server ..................................................... 87

5.16.14 Controller SHOW CLIENTS and MP Server Problem ............................... 88 5.16.15 Documentation Error – Record Streaming .................................................. 89 5.16.16 ResultSet.updateRow() and ResultSet.deleteRow() Problem ...................... 89 5.16.17 Problem using Column Renaming with dbkey ............................................ 89

5.16.18 Class cast error on Scaled integer Retrieval after ResultSet.insertRow() .... 90 5.16.19 ResultSet.deleteRow() Behaviour Change ................................................... 91

5.16.20 Underlying Blob Handles not Released when using ResultSet.getBlob() ... 92 5.16.21 Sequence Values not Visible to ResultSet get Methods using Column Name

................................................................................................................................... 93

5.16.22 Scaled Integer Problem ................................................................................ 95

5.16.23 Server Matching Exception may Stop Poll Handling .................................. 95 5.17 New Features for Release 7.3.0.0.x ...................................................................... 96

5.17.1 Shutdown Thread ........................................................................................... 96

5.17.2 Driver.attach() ................................................................................................ 96 5.17.3 Returning List of Known Databases .............................................................. 97

5.17.4 Controller Enhancements ............................................................................... 97 5.17.5 RDB_EXT.JAR file ....................................................................................... 97 5.17.6 Performance Enhancements ........................................................................... 99

5.18 Corrections in Release 7.3.0.0.x ......................................................................... 100 5.18.1 Server Startup Failure when using CFG File ............................................... 100

5.18.2 AccessViolation on Disconnect when Inserting Blobs ................................ 101 5.18.3 PreparedStatement and Parameter Markers Known Problem now Resolved

................................................................................................................................. 101 5.18.4 Controller SHOW CLIENTS and MP Server Problem ............................... 102

5.18.5 Possible Memory Leak when Updating Blob Columns ............................... 103 5.18.6 Access Violation with Trace and Network Dump ....................................... 103 5.18.7 Named Input Parameters not Working with CallableStatements ................ 104

5.19 New Features for Release 7.2.5.5.x .................................................................... 105 5.20 Corrections in Release 7.2.5.5.x ......................................................................... 105

5.20.1 Long-running Query Holds up New Connections in MP Server ................. 105

5.21 New Features for Release 7.2.5.4.x .................................................................... 106

5.22 Corrections in Release 7.2.5.4.x ......................................................................... 106 5.22.1 Access Violation at Java_rdb_JNI_SetStrVal ............................................. 106 5.22.2 DCL Command Line Too Long ................................................................... 107

7

5.22.3 Access Violation during DriverManager.getConnection() when Database

Specification is Missing .......................................................................................... 107 5.22.4 Unaligned Memory Faults on IA64 ............................................................. 108 5.22.5 Read-Only Transactions not Enforced on Connection Switch .................... 108

5.22.6 Sockets not Correctly Closing on OpenVMS Clients causing Accumulation of

Mailboxes ................................................................................................................ 109 5.22.7 Cast problem when Converting String to Date/Time .................................. 109

5.23 New Features for Release 7.2.5.3.x .................................................................... 111 5.24 Corrections in Release 7.2.5.3.x ......................................................................... 111

5.24.1 Interaction of DatabaseMetaData methods with Blobs may Crash the Thin

Server ...................................................................................................................... 111

5.24.2 BigDecimal scaling Incorrect when used with PreparedStatement SetObject()

Methods................................................................................................................... 112 5.24.3 Connection.nativeSQL() method Throws Null Pointer Exception .............. 112 5.24.4 Calling Resultset.isLast() method May Change Transaction Behavior ....... 112

5.25 New Features for Release 7.2.5.2.x .................................................................... 113 5.25.1 DEC_KANJI and DEC_HANZI Support Enabled ...................................... 113

5.26 Corrections in Release 7.2.5.2.x ......................................................................... 113 5.26.1 ResultSet.getBigDecimal() not Working with System ResultSets .............. 113 5.26.2 Setting TraceLevel fails when using Hexadecimal Notation ....................... 114

5.26.3 Delimited Identifier Problem in AS clause of Select Statement .................. 114

5.26.4 Configuration file problem in "DEFAULT" Server Definition ................... 114 5.26.5 Pool Server May choose Incompatible Pooled Server when User Restriction

Enabled ................................................................................................................... 115

5.26.6 Potential Problem when Dumping SQLDA in Trace .................................. 116 5.26.7 Connection.getCatalog() Returns Wrong Value for Single Schema Databases

................................................................................................................................. 116 5.26.8 Potential Memory Leak with Views ............................................................ 117


5.27.1 SQLDA Dumping ........................................................................................ 117 5.27.2 failSAFE IP with Pool Servers .................................................................... 117

5.27.3 HandshakeTries and HandshakeWait on Multi-process Native Connections

................................................................................................................................. 118 5.27.4 Server Access Security Enhancements ........................................................ 118 5.27.5 Restriction on using Multiple Blob fields in Join now Removed ................ 118

5.28 Corrections in Release 7.2.5.1.x ......................................................................... 119 5.28.1 Incorrect Row Number Returned after ResultSet.getLast() call .................. 119 5.28.2 Pool Server Startup of Pooled Servers may fail When Persona is Used ..... 120 5.28.3 Last Column in Select List may be Inaccessible in Some Queries .............. 121 5.28.4 Abnormal Client Termination may Prevent Executor Re-use ..................... 121

5.28.5 Decimal Column Problem with Native Driver ............................................ 122

5.28.6 'EFN xx is not available' Message on Executor Startup .............................. 122

5.28.7 Extraneous log message during Auto-restart check by Pool Server ............ 123 5.28.8 Logfile not Correctly set for Servers Started Using the Controller ............. 123

5.29 New Features for Release 7.2.5.0.x .................................................................... 123 5.29.1 Persona ......................................................................................................... 123

8

5.30 Corrections in Release 7.2.5.0.x ......................................................................... 124 5.30.1 Incorrect SQLSRV_JDBC_SERVER_STARTUP72 Installed with V7.2-41

Oracle JDBC for Rdb Kit ........................................................................................ 124 5.30.2 Multi-process Server May Show Continuous DIO Activity Even When Idle

................................................................................................................................. 124 5.30.3 Client idleTimeout Does Not Work for Prestarted and Reused Executors .. 125 5.30.4 Syntax Error in Query Generated for DatabaseMetaData.getTables ........... 125 5.30.5 Show Clients in Controller may Crash Connected Thin Server .................. 126


5.31.1 Client and Server Timeout Feature .............................................................. 127

5.31.2 Executor Name Prefix .................................................................................. 127

5.32 Corrections in Release 7.2.4.1.x ......................................................................... 127 5.32.1 Release Notes Specify Incorrect Installation Directory for

RDBJDBCCFG.XML ............................................................................................. 127 5.32.2 Persona Not Handled Correctly by the Multi-process and Pool Servers ..... 128

5.32.3 Multi-process Server / Executor Handshake Timeout May Be Too Short on

Heavily Loaded Systems......................................................................................... 128

5.32.4 Problems with srv.idleTimeout and srv.bindTimout Configuration Variables

and Their Use with SSL servers.............................................................................. 129 5.32.5 IA64 Problem Causes Array Out of Bounds Exception When Handling String

Indexing .................................................................................................................. 130

5.32.6 Comments within SQL Text Not Handled Correctly .................................. 130 5.32.7 Prepared Statements May Cause a Memory Leak with Multi-process Servers

................................................................................................................................. 131

5.33 Corrections in Release 7.2.4.0.x ......................................................................... 132 5.33.1 Maximum Size of Single Data Row Increased to 65,272 Octets ................. 132

5.33.2 Another Connection Overlap Window Found with Pool Servers ................ 132 5.33.3 SSL Server Information Not Correctly Set from XML-Formatted

Configuration File ................................................................................................... 132

9

Preface

Purpose of This Manual

The Oracle JDBC for Rdb 7.3.4.1.0 release notes summarize new features,

corrections to software, restrictions, workarounds, and problems. They also include

new features and corrections provided in releases 7.2.4.0.0. through 7.3.4.0.x. These

release notes cover Oracle JDBC for Rdb for OpenVMS on both Alpha and Integrity

Servers.

Intended Audience

This document is intended for users responsible for:

● System management

● Database administration

● Application programming

Access to Oracle Support

Oracle customers have access to electronic support through My Oracle Support. For

information, visit http://www.oracle.com/pls/topic/lookup?ctx=acc&id=info or visit

http://www.oracle.com/pls/topic/lookup?ctx=acc&id=trs if you are hearing impaired.

Document Structure

This document consists of five chapters:

Chapter 1 Describes location of documents and installation directions.

Chapter 2 Describes new features and technical changes in this release.

Chapter 3 Describes corrected software errors in this release.

Chapter 4 Describes known problems, restrictions, and workarounds.

Chapter 5 Describes new features and corrected software errors in

releases 7.2.4.0.0 through 7.3.4.0.x.

Conventions

Oracle JDBC for Rdb is often referred to as JDBC.

http://www.oracle.com/pls/topic/lookup?ctx=acc&id=info

http://www.oracle.com/pls/topic/lookup?ctx=acc&id=trs

10

Hewlett-Packard Company is often referred to as HP.

The following conventions are used in this document:

word A lowercase word in a format example indicates a syntax

element that you supply.

[ ] Brackets enclose optional clauses from which you can choose

one or none.

{ } Braces enclose clauses from which you must choose one

alternative.

... A horizontal ellipsis means you can repeat the previous item.

.

.

.

A vertical ellipsis in an example means that information not

directly related to the example has been omitted.

Conventions in Code Examples

Code examples illustrate Java code, SQL or other command-line statements. They

are displayed in a monospace (fixed-width) font and separated from normal text as

shown in this example:

SELECT last_name FROM employees WHERE last_name = 'TOLIVER';

The text of exception messages that may be raised are also displayed using the same

convention.

▲contents

11

Chapter 1 Installation and

Documentation

This chapter contains installation and documentation information for Oracle JDBC

for Rdb 7.3.4.1.x.

1.1 Accessing the Documentation

You can extract release notes or an Oracle JDBC for Rdb document from the PCSI kit

prior to installation by following one of these procedures:

● To extract a copy of the release notes, define PCSI$SOURCE to point to the

location (device name and directory) of the PCSI kit. Then, enter the PRODUCT

EXTRACT RELEASE_NOTES command followed by the product name at the DCL

prompt.

$ DEFINE PCSI$SOURCE DKA400:[KITS]

$ PRODUCT EXTRACT RELEASE_NOTES RDBJDBC73

● To extract a list of files contained in a software product kit, define

PCSI$SOURCE to point to the location (device name and directory) of the PCSI kit.

Then, enter the PRODUCT LIST command followed by the product name at the DCL

prompt.


$ PRODUCT LIST RDBJDBC73

● To extract a specified file, define PCSI$SOURCE to point to the location (device

name and directory) of the PCSI kit. Then, enter the PRODUCT EXTRACT FILE

command followed by the product name and file name at the DCL prompt.


$ PRODUCT EXTRACT FILE RDBJDBC73/SELECT=filename.ext

The Oracle JDBC for Rdb documentation is also available on My Oracle Support and

OTN.

The installation procedure copies the Oracle JDBC for Rdb release notes to the

SYS$HELP directory.

12

1.2 System and Software Requirements

Oracle JDBC for Rdb requires the following software products to be installed:

Software Minimum Version

Alpha Integrity

HP OpenVMS V8.2 V8.2-1

HP Java tm

SDK/RTE V5.0-1

(1.5.0-1)

V5.0-1

(1.5.0-1)

Oracle Rdb V7.2.1 V7.2.1

On the client side, you must install the following software product in order to use the

Oracle JDBC for Rdb Thin driver:


Java tm

SDK/RTE V1.5.0-1

In addition, if you need to start and stop Oracle JDBC for Rdb servers using Oracle

SQL/Services, the following product must be installed:


Alpha Integrity

Oracle

SQL/Services V7.1.6 V7.2

Detailed information about installing Hewlett-Packard Java for OpenVMS system

may be found at the following web site:

http://www.hp.com/java.

Documentation for HP's Java for OpenVMS system may be found at the following

web sites:

http://h18012.www1.hp.com/java/documentation/index.html

http://www.hp.com/java

http://h18012.www1.hp.com/java/documentation/index.html

13

In line with HP recommendations for Java applications, Oracle recommends the

following minimum quota setting on accounts used to start up Thin servers, in

particular those used to start Multi-process servers.

UAF Fillm 4096

Channelcnt 4096

Wsdef 4096

Wsquo 8192

Wsextent and Wsmax 32768

Pgflquo 3145728

bytlm 4000000

biolm 150

diolm 150

tqelm 100

Be sure to set your systems quotas appropriately to accommodate these process

quotas.

See the Java for OpenVMS release notes for more information on OpenVMS quotas

and resources required by Java.

Also refer to your Oracle Rdb documentation for recommendations on OpenVMS

quotas required for Oracle Rdb.

1.3 Installation

This section describes how to install Oracle JDBC for Rdb and includes a sample log.

1.3.1 Remove Prior BETA Versions of Oracle JDBC for Rdb Release 7.3.x.x.x

If you were participating in a BETA trial of Oracle JDBC for Rdb 7.3.x.x.x then you

must remove any previously installed BETA kits before proceeding with the

installation of this release.

Failure to remove previous 7.3 kits may prevent Oracle JDBC for Rdb from

functioning correctly. In particular, several changes have been made to the

client/server communication protocol during the life of the 7.3 BETA releases that

may cause connection failures if 7.3 BETA thin driver jars are used to connect to a

7.3 server or vice-versa.

14

In addition, ensure that all deployed copies of the 7.3.x.x.x BETA rdbThin.jar are

replaced. If not, connection failures may result in the applications using these BETA

driver jars.

1.3.2 Contents of the Oracle JDBC for Rdb Kit

The Oracle JDBC for Rdb kit uses OpenVMS Polycenter to simplify the installation

of the product. Please refer to your OpenVMS documentation on the use of

OpenVMS Polycenter.

The Oracle JDBC for Rdb kit product installation file is named ORCL-pppVMS-

RDBJDBC73-V0703-xxxxxx-1.PCSI where ppp will be the platform and xxxxxx

will be the build instance of this kit, for example:

ORCL-AXPVMS-RDBJDBC73-V0703-0V0A18-1.PCSI

or ORCL-I64VMS-RDBJDBC73-V0703-0V0A18-1.PCSI

The installation file is located in the RDBJDBC directory of the Rdb Software

distribution CD. If you obtained the Oracle JDBC for Rdb kit from the Web, the

installation file is contained in the RDBJDBCV73xxxxx.ZIP file, where xxxxx refers

to the build instance of the kit.

The installation kit is comprised of the following files:

BUILD_CERTS.COM

Command procedure example

of building certificates for

SSL.

OracleJDBCForRdbClient73000.msi

OracleJDBCForRdbClient73000X64.msi

Microsoft installer file for

client-side software

installation on Windows.

RDBJDBCCHECKUP.CLASS Use to verify the installation

of this kit.

RDBJDBCCHECKUP.JAVA Use to verify the installation

of this kit.

RDBJDBCEXEC73.EXE Executor image in conjunction

with a Multi-process server.

RDBJDBCCFG.XML Example XML formatted

configuration file.

RDBJDBCSHR73.EXE Shared image required for

Oracle Rdb database access.

RDBJDBCMPSHR73.EXE Shared image required for

15

Multi-process Oracle Rdb

database access.

RDBJDBC_EXECCLI.COM CLI helper command

procedure.

RDBJDBC_INSTALL.COM

Installation command

procedure used by Polycenter

during installation.

RDBJDBC_STARTEXEC.COM

Command procedure used by

Oracle JDBC for Rdb Multi-

process server to start up an

executor process.

RDBJDBC_STARTSRV.COM

Command procedure used

when Oracle JDBC for Rdb

servers are started up from the

Oracle JDBC for Rdb

controller.

RDB_EXT.JAR

Java jar file containing classes

and java code for extensions

to Oracle JDBC for Rdb.

RDBNATIVE.JAR

Java jar file containing the

classes for the Oracle JDBC

for Rdb native driver.

RDBTHIN.JAR



for Rdb thin driver.

RDBTHINSRV.JAR



for Rdb servers.

RDBTHINCONTROL.JAR



for Rdb controller.

RDBTHINSRVPOOL.JAR



for Rdb pool server.

RDBMANSRV.JAR



for Rdb manager server.

RDBJDBC_<version>_RELNOTES.PDF Oracle JDBC for Rdb Release

Notes.

RDBJDBC_<version>.RELEASE_NOTES Text version of Oracle JDBC

for Rdb release notes.

16

RDBJDBC_USERGUIDE.PDF User guide.

SQLSRV_JDBC_SERVER_STARTUP*.COM

Command procedures used by

Oracle SQL/Services to start

up an Oracle JDBC for Rdb

server.

WATCHEVENTS.JAR

Sample application for

watching Oracle JDBC for

Rdb server events.

Note

The HTML version of the user guide and release notes are no longer supplied.

1.3.3 Installation Procedure

Follow these steps to install the Oracle JDBC for Rdb kit:

1. If you obtained the kit in ZIP format, restore the kit file to a temporary directory:

$ unzip RDBJDBCV73xxxx.ZIP -d MY_DIR

This will unzip the Polycenter kit for Oracle JDBC for Rdb and you will have access

to the PCSI file, ORCL-pppVMS-RDBJDBC73-V0703-xxxxxx-1.PCSI, where ppp is

the platform and xxxxxx is the build instance of this kit.

2. Use the Polycenter PRODUCT command to install the kit.

Details of the version of the kit will be displayed and you will be asked if you want to

proceed. The following examples of installation on an ALPHA system assume the kit

build instance is 0V0A18, and that the directory where the PCSI file can be found is

MY_DIR.

$ PRODUCT INSTALL RDBJDBC73/SOURCE=MY_DIR

The following product has been selected:

ORCL AXPVMS RDBJDBC73 V7.3-0V0A18 Layered Product

[Installed]

Do you want to continue? [YES]

Configuration phase starting ...

17

You will be asked to choose options, if any, for each selected

product and for any products that may be installed to satisfy

software dependency requirements.

ORCL AXPVMS RDBJDBC73 V7.3-0V0A18: Oracle JDBC for Rdb

Copyright © 1995, 2010,Oracle Corporation. All Rights Reserved.

This product does not have any configuration options.

Execution phase starting ...

The following product will be installed to destination:

ORCL AXPVMS RDBJDBC73 V7.3-0V0A18

DISK$AXPVMSSYS:[VMS$COMMON.]

Portion done: 0%...10%...30%...40%...50%...60%...70%...80%...90%

Oracle JDBC for Rdb has been successfully installed in :

DISK$AXPVMSSYS:[SYS1.SYSCOMMON.rdb$jdbc.0703-0V0A18]

To help you setup the required logical names, a file named

RDBJDBC_STARTUP.COM has been added to this installation directory

RDBJDBC_STARTUP.COM:

$! Oracle JDBC for Rdb startup command procedure

$!

$ DEFINE/SYSTEM RDB$JDBC_HOME -

DISK$AXPVMSSYS:[SYS1.SYSCOMMON.rdb$jdbc.0703-0V0A18]

$ DEFINE/SYSTEM RDB$JDBC_LOGS -

DISK$AXPVMSSYS:[SYS1.SYSCOMMON.rdb$jdbc.logs]

$ DEFINE/SYSTEM RDB$JDBC_COM -

DISK$AXPVMSSYS:[SYS1.SYSCOMMON.rdb$jdbc.com]

$ DEFINE/SYSTEM RDBJDBCSHR RDB$JDBC_HOME:RDBJDBCSHR73.EXE

$ DEFINE/SYSTEM RDBJDBCMPSHR RDB$JDBC_HOME:RDBJDBCMPSHR73.EXE

$ DEFINE/SYSTEM RDBJDBCEXEC RDB$JDBC_HOME:RDBJDBCEXEC73.EXE

...100%

The following product has been installed:

ORCL AXPVMS RDBJDBC73 V7.3-0V0A18 Layered Product

The installation procedure will copy all the kit files to the appropriate Oracle JDBC

for Rdb product directory in the SYS$COMMON:[RDB$JDBC] directory, for

example:

SYS$COMMON:[RDB$JDBC.0703-0V0A18]

18

If they are not already present, the installation procedure will create two new

directories, SYS$COMMON:[RDB$JDBC.LOGS] and SYS$COMMON:[RDB$JDBC.COM].

Note

During installation a generic configuration file RDBJDBCCFG_TEMPLATE.XML

will be copied to the RDB$JDBC_HOME directory. Also during installation, if the

SYS$COMMON:[RDB$JDBC.COM]directory does not already contain a file named

RDBJDBCCFG.XML, the contents of the configuration template file will be used to

create this file.

By default, as the logical name RDB$JDBC_COM points to the

SYS$COMMON:[RDB$JDBC.COM] directory, it is this directory that the Oracle

SQL/Services JDBC dispatcher will use when searching for a configuration file to

use during server startup. ( See the sections JDBC Dispatcher and Determining the

server configuration file in the Oracle JDBC for Rdb User guide for more details).

The RDBJDBCCFG_TEMPLATE.XML found in the RDB$JDBC_HOME directory

will be replaced each time you install Oracle JDBC for Rdb, however, any existing

RDBJDBCCFG.XML file found in the RDB$JDBC_COM directory will not be

replaced. Oracle recommends to use

RDB$JDBC_HOME:RDBJDBCCFG_TEMPLATE.XML only as a template file and

not to use this file in production.

$ dir sys$common:[rdb$jdbc]/col=1

Directory SYS$COMMON:[RDB$JDBC]

0703-0V0A18.DIR;1

COM.DIR;1

LOGS.DIR;1

In addition, the command procedures SQLSRV_JDBC_SERVER_STARTUP*.COM

will be copied to the system specific SYS$MANAGER directory.

3. Use the command procedure RDBJDBC_STARTUP.COM found in the Oracle

JDBC for Rdb product installation directory to define the required system logical

names:

RDB$JDBC_HOME to point to the installation home.

$ define/system RDB$JDBC_HOME SYS$COMMON:[RDB$JDBC.0703-0V0A18]

19

RDB$JDBC_LOGS to point to the Oracle JDBC for Rdb log directory.

$ define/system RDB$JDBC_LOGS SYS$COMMON:[RDB$JDBC.LOGS]

RDB$JDBC_COM to point to the Oracle JDBC for Rdb command directory.

$ define/system RDB$JDBC_COM SYS$COMMON:[RDB$JDBC.COM]

RDBJDBCSHR to point to the shared image RDBJDBCSHR73.EXE.

$ define/system RDBJDBCSHR –

SYS$COMMON:[RDB$JDBC.0703-0V0A18]RDBJDBCSHR73.EXE

RDBJDBCMPSHR to point to the shared image RDBJDBCMPSHR73.EXE.

$ define/system RDBJDBCMPSHR –

SYS$COMMON:[RDB$JDBC.0703-0V0A18]RDBJDBCMPSHR73.EXE

RDBJDBCEXEC to point to the shared image RDBJDBCEXEC73.EXE.

$ define/system RDBJDBCEXEC –

SYS$COMMON:[RDB$JDBC.0703-0V0A18]RDBJDBCEXEC73.EXE

You must define the RDB$JDBC_HOME logical name if you want to use a Thin

Multi-process server or use the Controller or Pool servers or the SQL/Services JDBC

dispatcher to start server processes.

4. Include the rdbnative.jar and rdbthin.jar files in your Java CLASSPATH by using

either the logical names CLASSPATH or JAVA$CLASSPATH or the -classpath

option on the Java command line:

$ define JAVA$CLASSPATH -

[],RDB$JDBC_HOME:RDBNATIVE.JAR,RDB$JDBC_HOME:RDBTHIN.JAR

5. Test your installation using the "RdbJdbcCheckup" Java class in the

RDBJDBCCHECKUP.CLASS file. During the installation

RDBJDBCCHECKUP.CLASS is copied to RDB$JDBC_HOME.

Copy this file to your default directory and then you can invoke it using Java.

You will be prompted for a username and password and an Oracle Rdb database to

test the installation against. If the test succeeds, the text "Your JDBC installation is

correct." is displayed.

20

$ java "RdbJdbcCheckup"

Please enter information to test connection to the database

user: my_username

password: my_password

database: my_db_dir:personnel

Connecting to the database...Connecting

connected.

Hello World

Your JDBC installation is correct.

$

Test the Thin server by using the following commands:

$spawn/nowait/proc=rdbthinsrvtest java -jar rdbthinsrv.jar

$java "RdbJdbcCheckup" "-t"

Please enter information to test connection to the database

user: my_username

password: my_password

database: my_db_dir:personnel

Connecting to the database...Connecting...

connected.

Hello World

Your JDBC installation is correct.

$stop rdbthinsrvtest

Note

As Java is a case-sensitive language, it is important to specify class and

method names exactly as they are described in the various APIs. By

default, the OpenVMS operating system uppercases command line

parameters unless you surround them with double quotation marks.

1.3.4 Rdb Thin driver software on Windows

If you are developing or deploying Windows applications that use the Oracle JDBC

for Rdb thin driver, you will have to copy the provided RDBTHIN.JAR file to each

of the Windows systems you will be using. In addition you should consult your

application, IDE or Microsoft Windows software guides to determine how to let the

application or development environment know where the Rdb thin driver jar can be

found.

In order to facilitate installation of the Rdb thin driver on your Windows client

systems, included in the Oracle JDBC for Rdb installation kit are two Microsoft

Software Installation images, OracleJDBCForRdbClient73000.msi and

OracleJDBCForRdbClient73000X64.msi that may be used on your client Windows

systems to install the RdbThin and the RdbThinControl jars.

21

The OracleJDBCForRdbClient73000X64.msi may be used on Windows 64 Bit

systems. OracleJDBCForRdbClient73000.msi can also be used on Windows 64 Bit

systems; however the installation will be placed in the 32 bit subsystem.

Once the appropriate MSI file has been copied to your Windows system, double-click

the filename to invoke the Microsoft installer, and then follow the directions provided

by the installation Wizard.

This installation will place a copy of the RdbThin and RdbThinControl jars as well as

the current JDBC documentation into the designated installation directory on your

Windows system.

In addition, pointers to the uninstall procedure, the Rdb Thin Controller shortcut and

documentation will be added to your Program Menu.

1.3.5 Rdb Thin driver software on Other operating systems

If you are developing or deploying applications that use the Oracle JDBC for Rdb

thin driver, you will have to copy the provided RdbThin.jar file to each of the

systems you will be using.

You should consult your application, IDE or operating system software guides to

determine how to let the application or development environment know where the

Rdb thin driver jar can be found.

▲contents

22

Chapter 2 Enhancements Provided in

Oracle JDBC for Rdb Release 7.3.4.1.x.

This chapter describes new and changed features in Oracle JDBC for Rdb release

7.3.4.1.x.

2.1 New Transaction Option AUTOFETCH Release 7.3.4.1.0.

Also available in Release 7.3.4.0.4.

When AUTOCOMMIT is enabled, JDBC will automatically commit transactions at

the end of the execution of a statement.

Sometimes during long-duration client-side read operations it may be advantageous

to commit the current transaction more frequently, to prevent the read-only

transaction from locking out other database operations.

For example, when using SQLDeveloper to read through Rdb records using Oracle

JDBC for Rdb, records are delivered to the client application in groups, where the

number of records in a record group depends on the FETCHSIZE specified for the

statement or session. As you scroll down a list of records, the application may

request the next group of records from the server. This may mean that if you have

not finished scrolling through all the records in the retrieval set, there may still be a

transaction outstanding on the server waiting for the rest of the records to be

requested. If you leave this browse window open for an extended duration, this

could cause Read-Only transaction lockouts on the database.

JDBC now allows you to specify that transactions should be auto-committed each

time the server sends a FETCHSIZE number of records back to the client. This

means that while the client is browsing the current group of records, that no

transaction will be outstanding on the server.

A new connection string option “@TRANSACTION=AUTOFETCH” may be used

to inform JDBC that auto-commit is required each time the server sends

FETCHSIZE groups of records.

23

In addition the server configuration option “TRANSACTION=AUTOFETCH” is

also available to set the default transaction action for all connections that are made to

that server.

A second connection string option “@TRANSACTION=AUTOREADFETCH” is

also available and can be used to inform JDBC that, like the AUTOFETCH option,

auto-commit is required each time the server sends FETCHSIZE groups of records,

but when establishing a new transaction, if not otherwise explicitly specified, a Read-

Only transaction should be started.

See your JDBC documentation on FETCHSIZE and how to specify the number of

records to retrieve during client/server fetch operations. The application you are

using may also have documentation on how to specify the FETCHSIZE when using

JDBC drivers.

▲contents

24

Chapter 3 Problems Corrected

This chapter describes software errors corrected in Oracle JDBC for Rdb release

7.3.4.1.x.

3.1 Pool Servers Not Working

Fixed in Instance Build 20160205 release 7.3.4.1.0.

Also fixed in Instance Build 20150626 release 7.3.4.0.2.

A problem introduced in release 7.3.4.0.0 prevents pool servers from correctly

passing on client connection requests to the chosen pooled server.

The client application may receive an EOFException when trying to connect to

Rdb using a pool server.

This has now been fixed.

3.2 Problem using Multiprocess Connection Option

with the Native JDBC driver



A problem introduced in release 7.3.4.0.0 prevents the use of the @multiProcess

connection option when used in conjunction with the JDBC Native driver.

The following exception will be raised when the multiprocess switch is used:

java.lang.NullPointerException

at oracle.rdb.jdbc.util.Executor.runFromJAVA(Executor.java:155)

at oracle.rdb.jdbc.common.NativeConnect.connect(NativeConnect.java:284)

at oracle.rdb.jdbc.common.NativeConnect.connect(NativeConnect.java:94)


3.3 Bumpy-case Connection String Options used with

Properties.

25



As documented in the JDBC User Guide, some connection string options may

contain both upper and lower cased characters in their names (see section Connection

Options in the Oracle JDBC for Rdb User Guide).

A problem exists when using an option that has at least one upper-cased character in

its name. If the option is used within a Properties object passed to the

DriverManager.getConnection() method, JDBC may fail to recognized the option

and silently ignore it. For example, the option multiProcess used within a JDBC

Native driver connection may fail to be recognized if used in Properties:

String personnel="jdbc:rdbNative:personnel$db";

Properties props=new Properties();

props.put("user",user);

props.put("password",pass);

props.put("multiProcess","true");

Connection conn=DriverManager.getConnection(personnel, props);

A work around is to use the option directly on the URL connection string:

String personnel="jdbc:rdbNative:personnel$db@multiProcess=true"; Connection conn=DriverManager.getConnection(personnel,

user,pass);

3.4 Further Blob problems with Delimited Identifiers



The problem as discussed in release note 3.4 Blob Problem with Delimited Identifiers

above was not completely solved. It is possible that problems may still exist when

blobs take part in an Insert statement.

In particular if the table name associated with a segmented string is referenced in the

text of an INSERT statement is not fully uppercased and no delimited identifiers are

used in the statement then JDBC will fail with a syntax error:

oracle.rdb.jdbc.common.RdbException: Invalid column name

26

In addition, due to a problem introduced in release 7.3.4.0.0. and release 7.3.3.2.5. it

is possible that simple selection of segmented strings from rdb tables may fail with

an exception similar to:

%SQL-F-RELNOTDEF, Table xxxxxxx is not defined in database or schema

These problems have now been fixed.

3.5 Access Violation at

RDBJDBCMPSHR73+00227E31

BUG 22233633



During the run down of executors by the MP Server it is possible that an internal

error in memory handling may cause an Access Violation within the MP server

process similar to the following:

%SYSTEM-F-ACCVIO, access violation, reason mask=04,

virtual address=00000000000000CA, PC=FFFFFFFF904ED491,

PS=0000001B

If a stack trace is present or a system dump analysis of the errant process is done, it

may show reference to a PC at or near the following location:

RDBJDBCMPSHR73+00227E31

(This PC offset may differ depending on the version of JDBC installed; the value

above was seen when running Oracle JDBC for Rdb release 7.3.4.0. kit 0703-

4V0F6P)

This problem appears to occur only on Itanium systems and may occur when the

server node is very busy and is more likely to occur during the time an executor

process is being run-down.

Although this is the primary access violation, due to an error in the exception frame

handling by JAVA, the server process may also encounter a rapid iteration of access

violations similar to:

%SYSTEM-F-ACCVIO, access violation, reason mask=04, virtual

address=00000000000000CA, PC=FFFFFFFF904ED491, PS=0000001B


address=0000000000000000, PC=FFFFFFFF806BC5D0, PS=0000001B

27


address=0300B7590200BB2A, PC=00000000011D9820, PS=0000001B





.

.

.

HP OpenVMS engineering are aware of this JAVA exception handling looping

problem and are currently investigating it.

▲contents

28

Chapter 4 Known Problems and Workarounds

This chapter describes known problems, restrictions, and workarounds for Oracle

JDBC for Rdb Release 7.3.4.x.x.

4.1 Thin Server Deadlocks

When using a thin server, the Rdb database environment and binds are within the

OpenVMS process context of the server process.

When multiple clients attach to database using the same server, each client is

allocated its own thread and has its own connection context within Rdb, but from the

Rdb engines viewpoint, all of these connections are held by a single process.

The locking behavior of Rdb for connections within a single process differs from that

of connections made using discrete OpenVMS processes. In particular when two

separate processes try to get the same lock, by default, the second process will wait

until the first releases the lock. In contrast when two connections within the same

process try to get the same lock, a deadlock may be raised.

Oracle recommends, when using thin servers, that the lockwait connection switch

or server option be used to prevent deadlocks. By choosing a lock wait duration

shorter that your system deadlock wait, and then choosing an appropriate maxtries

value which determines how many times the server will attempt to get the lock before

giving up, deadlocks may be prevented. In addition keeping your transaction duration

small, will reduce the time locks are held and thus allowing better concurrency.

See Lockwait and Maxtries in your Oracle JDBC for Rdb User Guide for more

information.

4.2 Using Java Fast VM on OpenVMS ALPHA

Using Java Fast VM on OpenVMS Alpha when starting thin servers may limit the

number of clients a single server may be able to handle concurrently. This is because

using Fast VM drastically reduces the amount of certain system memory that the

Oracle Rdb subsystem has access to.

The usual symptom of running out of memory due to this situation is when the server

process issues COSI-VASFULL errors.

29

Refer to the OpenVMS Java documentation on using Fast VM for suggestions on

how memory usage may be altered.

Heap size used by the Java VM is important in determining how much memory will

be pre-allocated by the Java VM. You can set the size of the heap using the -Xmx

option. By default, the Fast VM looks at your quota and the size of physical memory

on the system to decide how large a heap to give you. So if both are very large, you

may wind up with a larger heap than you really need. You can use -verbosegc on the

command line of the command used to start a server to see the current heap size.

Memory usage may also be altered by using the "-Xglobal" switch.

If the thin servers are getting COSI-VASFULL errors when Fast VM is enabled,

Oracle suggests trying the following switch settings as a first pass at rectifying the

problem.

$ java "-Xmx24m" "-Xglobal120m" -jar rdbthinsrv.jar

4.3 Using the Oracle SQL/Services Management GUI

and JDBC Dispatchers

The unsupported Oracle SQL/Services Management GUI does not recognize

dispatchers of the type JDBC. Unfortunately, this means that you will no longer be

able to use this GUI once a JDBC dispatcher has been defined.

Removing the JDBC dispatcher from your Oracle SQL/Services definitions will

alleviate this problem.

Note: The unsupported Oracle SQL/Services Management GUI has been replaced

with the newer JAVA-based Oracle Rdb Connectivity Management GUI (ORCM).

4.4 Blob Columns and Correlation Names

Due to the nature of the parsing carried out by the Oracle JDBC for Rdb drivers it is

required that all blob columns referenced from the second and subsequent tables in a

multi-table join must be qualified using correlation names as shown below:

Select ta.blob, tb.blob from table1 ta, table2 tb

where ta.name = tb.name

30

Failure to use a correlation name in conjunction with the blob column name may

result in SQL parsing errors when data is retrieved from the blob field as the drivers

do not have enough information to determine the correct table to access the blob data

from.

SQL-F-FLDNOTCRS, Column <blob col> was not found in the tables

in current scope

This limitation also means that the use of "*" in the select clause for a join across two

or more tables that include blob fields may also cause a similar SQL error.

4.5 Blob Columns and Update Statements

When the SQL statements compiled by the Oracle JDBC for Rdb drivers contain

reference to list of byte varying columns (Blobs) the drivers must create auxiliary

statements to carry out the required operations on the Blob columns within the

database.

These statement handle the preparation of list cursors required by Rdb to access the

underlying list of byte varying columns and use the dbkeys of the target rows to

establish the correct currency on the parent tables when using the list cursors.

To achieve this, code is added to any Select queries containing reference to Blob

columns in order to retrieve the dbkey of the parent table row. Code is also added to

Insert and Update statement to return the dbkey of the affected row.

If a Blob column is updated within an SQL Update statement, it is required that the

execution of the statement only updates a single row as only a single dbkey can be

returned to the drivers.

For example:

.

.

.

PreparedStatement prep = Conn.PrepareStatement(

"update resumes set resume = ? where employee_id = '91111'");

prep.setBinaryStream(1, bs, sss.length());

prep.execute();

.

.

.

If the resumes table contains more than one row satisfying the query selection criteria

then an exception will be raised by Rdb during the execute() :

31

%RDB-E-MULTIPLE_MATCH, record selection criteria should identify

only one record; more than one record found

4.6 External Procedures and Thin Server

Due to the way that Rdb carries out internal communication of handles between

different component during external procedure execution it is possible that concurrent

clients within the same Thin Server trying to use external procedures may see

exceptions related to the Rdb error BAD_DB_HANDLE.

Currently it is not possible to rectify this limitation within the Thin Server, as the

process model used internally by Rdb for using external procedures does not align

with how multi-threading has to be carried out within the Thin Server.

As the Multi-process Server (MP) uses a separate OpenVMS process for each

connection, it aligns better with the expected Rdb process model and consequently

external procedures may be used by concurrent clients.

Oracle suggests that if external procedures are expected to be executed by concurrent

clients, an MP Server should be used to handle these connections.

4.7 Limitations

4.7.1 Unsupported Methods

The following JDCB 2.0, JDBC 3.0 and JDK 1.4 methods are not currently

supported:

Blob.setBytes

Blob.setBinaryStream

Clob.setString

Clob.setAsciiStream

Clob.setCharacterStream

Clob.truncate

Connection.setSavepoint

32

Connection.rollback(savepoint)

Connection.releaseSavepoint

DatabaseMeteData.getSQLKeywords

PreparedStatement.setRef

PreparedStatement.setArray

PreparedStatement.setNull(int,int,String)

PreparedStatement.setURL(int,URL)

ResultSet.getRef

ResultSet.getArray

ResultSet.updateAsciiStream

ResultSet.updateBinaryStream

ResultSet.updateCharacterStream

ResultSet.updateRef

ResultSet.updateArray

ResultSet.rowUpdated

ResultSet.rowInserted

ResultSet.rowDeleted

ResultSet.updateBytes

Statement.cancel

Statement.setQueryTimeout

Statement.getMoreResults

Statement.executeUpdate(String sql, int autoGeneratedKeys)

Statement.executeUpdate(String sql, int[] columnIndexes)

Statement.executeUpdate(String sql, String[] columnNames)

Statement.execute(String sql, int autoGeneratedKeys)

Statement.execute(String sql, int[] columnIndexes)

33

Statement.execute(String sql, String[] columnNames)

The following features or datatypes in JDBC 2.0 and JDBC 3.0 are not supported:

● Array

● Ref

● Clob

● User Defined datatypes

● Scroll cursors

● Savepoints

4.7.2 Auto-generated keys

The total number of markers and fields allowed in a single SQL statement is 250.

4.7.3 String Truncation Warnings

The Oracle JDBC for Rdb drivers follow the SQL-92 rules for string truncation that

differ depending on whether it is a store or retrieval.

If a string truncation happens during a store operation, Oracle Rdb signals the error

RDB$_TRUN_STORE, unless all of the truncated characters are spaces, in which

case there is no error. If a string truncation happens during a retrieval, Oracle Rdb

signals the SQL warning RDMS$K_SQLCODE_TRUNCWARN.

4.7.4 Numeric and String Functions in JDBC

A number of JDBC standard Numeric and String functions are not supported within Oracle

Rdb unless you have previously prepared the database for use with OCI Services for Oracle

Rdb using the sql_functions.sql script. Refer to the Oracle SQL/Services documentation for

more details on using this script.

▲contents

34

Chapter 5 New Features and Corrections in

Previous Releases

5.1 New Features for Release 7.3.4.0.0

This section describes new and changed features in Oracle JDBC for Rdb 7.3.4.0.0.

5.1.1 Create / Drop Database Support

Release 7.3.4.0.0.

Starting with release 7.3.4.0.0 of Oracle JDBC for Rdb, the JDBC drivers and servers

support the ability to create and drop databases.

JDBC will accept the standard Oracle Rdb SQL syntax for create and drop

database, however, syntactical keywords must not be abbreviated.

In addition the statement may also contain the clauses “IF EXISTS” or “IF NOT

EXISTS” to conditionalize the operation.

By default, JDBC servers will not allow CREATE or DROP database statements, the

server configuration must be changed to enabled these operations.

See the Create and Drop Database section in your Oracle JDBC for Rdb User Guide

for more information.

▲contents

5.2 Corrections in Release 7.3.4.0.0

This section describes software errors corrected in Oracle JDBC for Rdb 7.3.4.0.0

(Also see section Problems Corrected in release 7.3.4.1.x for details of any fixes that

may be included in patch releases 7.3.4.0.2 through 7.3.4.0.5).

35

5.2.1 Like Escape Throws OutOfBounds Exception

Fixed in Instance Build 20150424 Release 7.3.4.0.0.

Also fixed in release 7.3.3.2.2.

The use of the escape clause within a SQL statement containing the like operator

may cause the JDBC driver to throw an OutOfBounds exception.

The problem only occurs if the escape character chosen is the same default escape

character used by JAVA in string literals, the backslash character “\”, for example:

select 'not ok' from rdb$database where 'a%a' like '%\%%' escape '\'

A workaround is to use a character other than backslash for the escape character:

select 'ok' from rdb$database where 'a%a' like '%|%%' escape '|'


5.2.2 Correlation Name Workaround for Blobs Fails



A problem introduced prior to Release 7.3.3.0.0 of Oracle JDBC for Rdb prevents

the workaround as described in the Known Problems and Workarounds section Blob

Columns and Correlation Names from working.


5.2.3 PreparedStatement Cache Failure



PreparedStatement handles can be cached by using the sqlcacheps connection

property (See Caching Statement Handles in the Oracle JDBC for Rdb User

Guide).

If statement handle caching has been enabled, a problem with return parameter

handling may cause a cached PreparedStatement to fail during execution with the

following exception:

oracle.rdb.jdbc.common.RdbException: Missing IN or OUT parameter at index: n

Where the index ‘n’ shown is a number 1 greater than the actual number of

parameter markers in the compiled statement.

36


5.2.4 Blob problem with Delimited Identifiers



A problem in the production of SQL statements during the processing of blobs

internally by JDBC may cause a SQL syntax exception to be raised.

When storing or updating blobs JDBC has to create SQL statements to handle the

Rdb segmented string operations. If the table name or column name associated with

the blob field requires to be handled as a delimited identifier, JDBC fails to

recognize this, and generates SQL text that is not syntactically correct.

▲contents

5.3 New Features for Release 7.3.3.2.x

This section describes new and changed features in Oracle JDBC for Rdb 7.3.3.2.x.

5.3.1 Change Transaction Start Time.

Build 20141215 release 7.3.3.2.2.

By default, when a new connection is made, JDBC will immediately start a

READ_ONLY transaction in anticipation of subsequent read-only operations.

The type of the default transaction may be changed to READ_WRITE using

connection string options.

The start of the default transaction may now be deferred by using the new connection

string transaction options, READONLYDEFER and READWRITEDEFER. See the

section Connection Options in the Oracle JDBC for Rdb User Guide for more

details.

▲contents


37

This section describes software errors corrected in Oracle JDBC for Rdb 7.3.3.2.0.

(Also see Corrections in Release 7.3.4.0.0 section for details of any fixes that may be

included in patch releases 7.3.3.2.2 through 7.3.3.2.5).

5.4.1 NullPointerException when allowing Anonymous

Fixed in Instance Build 20140905 release 7.3.3.2.0. Also fixed in release 7.3.3.1.2.

When anonymous access to a JDBC server has been enabled, the server should allow

clients to use a blank username in order to access the Rdb database anonymously.

A problem introduced during the addition of user restriction features will cause the

server to throw a NullPointerException, if the server has anonymous access enabled and

the client provides a blank username.


5.4.2 UnsupportedEncodingException: Binary



During the processing of Blob data from the database the Rdb thin driver may throw

the following exception:

java.io.UnsupportedEncodingException: BINARY

at java.lang.StringCoding.encode(StringCoding.java:269)

or

java.io.UnsupportedEncodingException: BINARY : BINARY at

oracle.rdb.jdbc.common.Column.getObjectFromDb(Column.java:1818


5.4.3 Alignment Faults when using MP Server

(BUG 20046411)



38

A fix made in release 7.3.3.1.0 to solve a problem with MUTEX handling within the

MP Server has exposed a misaligned internal structure that may cause excessive

alignment faults on IA64 systems.

This problem only affects MP servers.


5.4.4 ResultSet.absolute() Problems

(BUG 19780721)

Fixed in Instance Build 20141121 release 7.3.3.2.0. Also fixed in release 7.3.3.1.6.

Several problems have been found when using the ResultSet.absolute() method

when specifying negative row numbers or row numbers that are greater than the number

of rows returned by the ResultSet.

An error similar to the following may be seen:

SQL Exception Connection lost : n >= n

In addition the absolute() method may return “TRUE” when no row is available.

5.4.5 Orphaned Executor Processes when using Persona

(BUG 20119695)


A problem with lost executor checking within the MP server may cause the server to

prematurely attempt to terminate viable executors.

The problem is seen only when the MP server is started using the “-persona”

configuration option.

A problem in the termination protocol may also prevent the correct termination of the

executor process. As the executor process is not terminated fully, resources may be held,

which may cause resource problems if a number of these orphaned executor processes

exist concurrently on your system.

A workaround to prevent the premature executor termination is to disable the lost

executor checking by setting the MP Server configuration option

“-srv.lostexecutorsweep” to the value “0”.

Another possible work-round is to not use persona when using an MP server.

39

Standard thin servers using persona are not affected.

▲contents

5.5 New Features for Release 7.3.3.1.0

This section describes new and changed features in Oracle JDBC for Rdb 7.3.3.1.0.

5.5.1 Change in Template Configuration Filename

Release 7.3.3.1.0.

In prior releases the JDBC installation would place a copy of the template file

RDBJDBCCFG.XML in the RDB$JDBC_HOME directory and if not already

present, a copy will be placed in the RDB$JDBC_COM directory.

The copy in RDB$JDBC_HOME is meant to be a reference template copy only and

should not be used directly in production. In order to reduce possible confusion about

the use of this file, starting with JDBC release 7.3.3.1.0, the copy installed to

RDB$JDBC_HOME is renamed to RDBJDBCCFG_TEMPLATE.XML.

The RDBJDBCCFG_TEMPLATE.XML found in the RDB$JDBC_HOME directory

will be replaced each time you install Oracle JDBC for Rdb, however, any existing

RDBJDBCCFG.XML file found in the RDB$JDBC_COM directory will not be

replaced.

Note: By default, as the logical name RDB$JDBC_COM points to the

SYS$COMMON:[RDB$JDBC.COM] directory, it is this directory that the Oracle

SQL/Services JDBC dispatcher will use when searching for a configuration file to

use during server startup. ( See the sections JDBC Dispatcher and Determining the

server configuration file in the Oracle JDBC for Rdb User guide for more details).

▲contents

40


This section describes software errors corrected in Oracle JDBC for Rdb 7.3.3.1.0.

(See section Corrections in Release 7.3.3.2.x for details of any fixes that may be

included in patch releases 7.3.3.1.2 through 7.3.3.1.6).

5.6.1 DatabaseMetadata.getColumns Problem with DefaultValue

(BUG 18900456 ) Fixed in Instance Build 20140717 release 7.3.3.1.0.

The DatabaseMetadata class method getColumns should return the default value

for each column found for the table selection specified.

A problem introduced in Oracle JDBC for Rdb Release 7.3.3.0 in the handling of

internal segmented string columns used to derive the underlying default value may

cause JDBC to either indicate that there is no default value for the column when one

actually exists, or may throw the following exception:

java.io.UnsupportedEncodingException: BLR : BLR

JDBC now correctly returns the default value found for either the table column or the

domain that the column is based on.

5.6.2 Show Server Configuration file Problem


When the Show Server command is used within the Oracle JDBC for Rdb

controller, details about the connected server are displayed. Included in the standard

display is the name of the configuration file the server used on startup.

In prior versions, JDBC used the standard Java File.getCanonicalPath() method to

retrieve the current configuration file specification. However on OpenVMS, if the

file specification contains a logical name and depending on whether the logical name

pointes to a rooted directory, the canonical form may contain extraneous path

elements within the same path string, for example:

SYS$COMMON:[rdb$jdbc.0703-3v0e5s]rdb$jdbc_com:rdbjdbccfg.xml

As shown in the above example, the path may display both the logical name and the

translation of that name.

41

Show Server has now been changed to no longer use the canonical form to display

the configuration file specification.

5.6.3 Unknown Action Message in Log file


Starting with Oracle JDBC from Rdb Release 7.3.3.0.0 it is possible that the

following message text may be displayed in the server log file:

JNI_Action 10 <unknown ACTION>: 10

This action code relates to a new feature (executor checking) introduced in 7.3.3.0.0

and the log message displayed is due to a missing description string within the debug

output code of the server.

This message does not represent a true error and may be safely disregarded.

5.6.4 Severe Performance Degradation in MP Server

(BUG 19150908 )


A change in the handling of mutex operations within the Multiprocess server made

during the 7.3.3.0.0 release may cause a severe reduction in the performance of the

server seen especially when the server is being utilized on a very busy system.


5.6.5 Controller DCL Command line Failure

(BUG 18973965 )


A problem introduced in release 7.3.3.0.0 may cause the DCL command line

parameter –node (or –n) to be incorrectly parsed by the Oracle JDBC for Rdb

controller. DCL command line statements similar to the following used to invoke the

controller will fail:

42

$ java -jar rdb$jdbc_home:rdbthincontrol.jar –

–n localhost -p 1701 -showclients

The statement fails showing the following exception:

Exception: java.lang.NumberFormatException: For input string: "localhost"

A possible workaround is to use the –url command line option instead:

$ java -jar rdb$jdbc_home:rdbthincontrol.jar –

–url //localhost:1701/ -showclients

5.6.6 Server Access Restrictions Ignored (BUG 19149514 )


A problem in how restriction checks are parsed from the server configuration file

may cause the JDBC server to ignore certain restrictions.

For example the following restriction may be ignored by the server if there are no

other allow or deny restrictions specified for the server; the server will incorrectly

allow all IPs instead of only local node users:

<server

name = "RestrictServer"

type = "RdbThinSrv"

url = "//localhost:1701/"

restrictAccess = "true">

<allowIP IP = "127.0.0.1"/>

</server>

This problem only occurs when the older access restriction syntax is used, in

particular:

AllowIP IP=”…”

AllowUser name=”…”

A workaround is to use the newer access restriction syntax:

Allow IP=”…”

Allow User=”…”

For example:

43

<server

name = "RestrictServer"

type = "RdbThinSrv"


restrictAccess = "true">

<allow IP = "127.0.0.1"/>

</server>

5.6.7 Datetime Insertion Issues

(BUG 18921457)


Release 7.3.3.0.0 introduced the ability to maintain timestamp information at the

nanosecond level. This feature is only available if both the client and the server sides

of the JDBC connection are at Oracle JDBC for Rdb release 7.3.3.0.0 or higher. If

lower JDBC releases are used, timestamp handling will resort to the default

millisecond level.

Unfortunately the code required to ensure JDBC upward and downward release

compatibility introudced some problem in the handling of some datetime operations.

The PreparedStatement setObject() method may fail to insert the correct datetime

value into the Rdb database when the destination column is a datetime columns such

as Timestamp or VMS Date or Time and the source is a date/time object.

For example, the following may load the timestamp column with incorrect datetime

values:

Given the following table: date1 (f1 integer, ts timestamp(2))

String inTimestamp = "2014-07-21 09:22:10.1234567";

java.sql.Timestamp cvtTimestamp =

java.sql.Timestamp.valueOf(inTimestamp);

DateFormat dateFormat = new SimpleDateFormat("yyyy-MM-dd");

java.util.Date cvtDate = dateFormat.parse("2014-07-21");

java.sql.Date cvtsdate =

java.sql.Date.valueOf("2014-07-21");

java.sql.Time cvtstime = java.sql.Time.valueOf("09:22:10");

long cvtlong = cvtTimestamp.getTime();

PreparedStatement ps3 = conn.prepareStatement(

"insert into tsbug1 (f1,ts) values (?,?)");

ps3.setObject(1, 1);

ps3.setObject(2, cvtTimestamp);

ps3.executeUpdate();

44


ps3.setObject(2, cvtDate);



ps3.setObject(2, cvtsdate);



ps3.setObject(2, cvtstime);



ps3.setObject(2, cvtlong);


Each record created will have an incorrect datetime value stored in the database.

Note: When the setObject() source is a datatype that will naturally cast to a long

value, the long value derived will be considered to be a milliseconds value

representing the number of milliseconds that have passed since January 1, 1970

00:00:00.000 GMT.

A workaround is to use the appropriate datatyped SET method instead of the generic

setObject() method, for example:

ps3.setTimestamp(1, cvtTimestamp)

5.6.8 MP Server may Become CPU Bound


A problem introduced in 7.3.3.0.0 may cause the MultiProcess server to go into a

tight CPU loop and consume excessive CPU even when no clients are attached.

This problem is a result of an unhandled array list exception that may manifest as the

following execition that will be shown in the server log file:

java.util.ConcurrentModificationException


5.6.9 Controller Null Pointer Exception when no Default Server

45


When starting a server using the controller DCL command line option –

startserver, the name of the server provided is used to locate the server

characteristics within the configuration file used by the controller.

If the named server does not exist, the controller will then use the characteristics of

the server named “Default”. Due to a problem in the server configuration setup, if

the configuration file does not contain a Default server definition, a Null Pointer

exception is thrown.

The workaround for this is to either specify the name of a server that does exist

within the configuration file, or alternatively, ensure that the configuration file does

have a “Default” server specified, for example:

<?xml version = '1.0'?>



<config>



<servers>



<server

name="DEFAULT"

url="//localhost:1701/"

/>

<server

.

.

.

This problem has now been fixed. If the configuration file has no “Default” server

specified and the named server does not exist, the server will be started using the

default thin server port 1701.

5.6.10 Pool Server Tries to Restart Running Server


The pool server will attempt to restart any server in its pool that is marked for restart

and has been found to unresponsive. A problem in the viability check for servers

may lead the pool server to incorrectly mark a server as non-responsive even though

the server is running and responsive.

46

In releases prior to 7.3.3.0.0 the pool server will immediately remove servers from its

pool it thinks are unresponsive, thus preventing them from taking part in the pool.

The server will no longer be chosen as a candidate for new client connections.

In release 7.3.3.0.0 and above, the pool server will maintain the server in its pool but

will periodically check to see if it has become responsive again. As the server is still

in the pool it may still receive and process new connection requests from the pool

server, however after a predefined number of viability checks, if the pooled server is

still deemed unresponsive it will then be removed from the pool, and no further

connections will be redirected to it.

In addition, if the pool server finds an unresponsive server, and that server is marked

for restart, the pool server will attempt to restart this server, however if the server is

still running the restart attempt will fail and an error will be logged in the pool server

log.

A workaround for this problem is to change the pool server configuration file and set

all pooled servers autorestart to false, for example:

<server

name = "srv1"

type = "RdbThinSrv"


autostart = "true"

autorestart = "false"

/>

The pool server has now been fixed to improve its pool viability checking.

5.6.11 Controller Poll Reopenlogs Hangs


If the controller command POLL REOPENLOGS is issued as the first POLL

command within the controller session, the controller will hang.

A workaround is to issue a standard POLL command first prior to issuing the POLL

REOPENLOGS:

rdbthincontrol> poll

Polling servers ...

... Polling complete : 1 responded. RdbThinSrv1701(0) //192.1.32.212:1701/ (0x2542DAD9<625138393>)

node = alfred

47

rdbthincontrol> poll reopenlogs

RdbThinSrv1701(0) //192.1.32.212:1701/ (0x2542DAD9<625138393>)

node = alfred : Logs Reopened

rdbthincontrol>

5.6.12 Pool Server Usage Balancing Problem


When the server balancing attribute for a Pool Server is set to “USAGE”, the pool

server should choose the server from its pool that has the most free user slots

available. The free user slots count is the number of potential connections the server

can still allow before the server MAXCLIENTS is reached.

In addition, any server that currently has no client connections should be

preferentially chosen.

A problem in the usage algorithm may cause the pool server to disregard servers that

have no current connections and select the least used server from the list of servers

that have at least one current connection.

This has now been fixed, as documented in the Oracle JDBC for Rdb User guide, if a

server does not currently have any client connections, it will be chosen in preference

to other busy servers.

5.6.13 Configuration File Problem Regression


A problem with the inheritance of server attributes from the “DEFAULT” server was

fixed in release 7.3.2.0.0. This problem has unfortunately resurfaced.

The problem and workaround is described in release note:

5.4.3 Some Server Characteristics not Correctly Inherited from DEFAULT


5.6.14 Server Configuration Deny User not Enforced


48

The server attribute ‘deny user="xxxxx"’ in conjunction with the attribute

‘restrictAccess="true" ‘ may be used to deny specific users from accessing

a JDBC server:

<server

name = "srv1"

type = "RdbThinSrv"


restrictAccess = "true"

>

<deny user = "murray"/>

</server>

However, a problem in the configuration attribute parsing may prevent the server

from correctly denying access to the specified users.

A workaround for this problem is to also include a valid IP restriction for the same

server definition. The wildcard IP address may be used:

<server

name = "srv1"

type = "RdbThinSrv"



>

<deny user = "murray"/>

<allow ip = "*.*.*.*"/>

</server>

▲contents

5.7 New Features for Release 7.3.3.0.x.


5.7.1 Oracle Rdb PID now Displayed

Fixed in release 7.3.3.0.0.

Also available release 7.3.2.0.2.

49

The process/stream ID used internally by Oracle Rdb to uniquely identify client

connections and that is displayed in output from operations such as RMU/SHOW

USERS is now displayed by JDBC in server logs and in controller SHOW CLIENT

output.

The Rdb process/stream information is only available to JDBC if the client has

connected to a database within an Rdb environment of Version 7.2.5.0.0 or above,

and the JDBC server used is release 7.3.2.0.2 or above.

See the Show Clients section of the Oracle JDBC for Rdb User Guide for more

information.

5.7.2 Controller Command Show Executors


Also available release 7.3.2.0.2.

A new controller command show executors is now available to display

information about executors hosted by the currently connected multi-process server.

See the Showing Executors section of the Oracle JDBC for Rdb User Guide for more

information.

5.7.3 Extra Timestamp Precision


JDBC has now been enhanced to allow the use of extra precision in the fractions of

seconds associated with timestamps.

Prior to this release, when transferring data between Rdb datetime or timestamp

columns and Java timestamp variables, JDBC would limit the precision of second

fractions to 3 decimal places, i.e. milliseconds.

If the Rdb timestamp or datetime column was set by Rdb, for example when using

CURRENT_TIMESTAMP, the precision of the seconds value would be further

limited to the maximum precision supported by Rdb SQL , i.e. 2 decimal places (

hundreds of seconds).

It is possible to load higher precision time data into the timestamp columns

programmatically, however earlier version of JDBC would truncate the value to

milliseconds during datetime operations.

JDBC has now been changed and will now allow the use of timestamp data with upto

7 decimal places (i.e. OpenVMS ticks, or 100 nanosecond units).

50

The extra timestamp precision is only available to applications using JDBC 7.3.3.0

Patch 2 (or later) native or thin drivers, and only available to the thin driver if it is

connected to a JDBC server running release 7.3.3.0 Patch 2 (or later).

If required, the older precision of 3 decimal places for the fraction of seconds may

retained for individual database connections by using the “@TICKS” connection

string qualifier set to false.

For example:

Connection conn = DriverManager.getConnection(

"jdbc:rdbThin://bravo:1755/my_db_dir:pers.rdb@ticks=false",

user, pass);

See the Connection Options section of the Oracle JDBC for Rdb User Guide for

more information.

5.7.4 Nlslang Connection Switch


When the values of text columns or literals are moved between Rdb and the JDBC

internal buffers, JDBC must convert the value to and from Unicode. To do this

JDBC need to know the expected character set encoding that the Rdb data is using.

JDBC will derive the encoding required using the character set information supplied

to it by Rdb.

Historically, especially prior to the current character set support found in Rdb, non-

MCS characters such as Kanji characters may have been stored in columns or literals

that have the DEC_MCS character set associated with them. If this is the case, when

JDBC converts these values to Unicode the original character may be lost as JDBC

will assume that the character was DEC_MCS.

To provide a way for customers that have non-MCS characters stored in DEC_MCS

literals or columns to correctly pass the original characters to JDBC applications,

JDBC now provides a connection string option nlslang.

This option may be used to tell JDBC the actually encoding used in columns or

literals that have the DEC_MCS character set attribute.

For example:


"jdbc:rdbThin://bravo:1755/my_db_dir:pers.rdb@nlslang=DEC_KANJI",

51

user, pass);


more information.

Note: This option does not alter the SQL National Language.

5.7.5 Use Query Header as Description

Fixed in release7.3.3.0.2.

The method DatabaseMetadata.getColumns() provides information about

table columns to the calling JDBC client. One of the standard properties returned by

the getColumns() method is Remarks which JDBC will fill from the column

description field found for that column returned by Oracle Rdb. By default, JDBC

uses the RDB$DESCRIPTION field of the RDB$RELATION_FIELDS system

table for the specified column.

JDBC will now allow you to specify that the Remarks property should be derived

from the Query Header field (the RDB$QUERY_HEADER column) of the

RDB$RELATION_FIELDS instead of the RDB$DESCRIPTION column.

The new connection string qualifier @useQueryHeader may be used to tell

JDBC to use the alternate source of the column’s Remarks field. If the value is set to

true, JDBC will fill the Remarks field from the RDB$QUERY_HEADER column

of the RDB$RELATIONS_FIELDS table. For example:


"jdbc:rdbThin://bravo:1755/my_db_dir:pers.rdb@useQueryHeader=true",

user, pass);

Optionally, you may also specify a specific character set to use to encode the

contents of the query header column when returned to your application. For example:


"jdbc:rdbThin://bravo:1755/my_db_dir:pers.rdb"+

"@useQueryHeader=Dec_Kanji", user, pass);

If you do not specify a character set to use but have specified an alternate character

set using the “@NLSLANG” connection string qualifier, JDBC will use that character

set to carry out appropriate encoding of the query header.

52

If no character set is specified in the “@useQueryHeader” connection string

qualifier or specified by the “@NLSLANG” connection string qualifier then LATIN1

will be used during encoding.


more information.

▲contents

5.8 Corrections in Release 7.3.3.0.x

This section describes software errors corrected in Oracle JDBC for Rdb 7.3.3.0.x.

5.8.1 Controller START SERVER Problems


The controller will fail with a Java NullPointer exception if the Start Server

command is used within a controller session that does not have a configuration file

specified:

$ java -jar RDB$JDBC_HOME:RDBTHINCONTROL.JAR

rdbthincontrol> start server 1701

NullPointerException:


A work around for this problem is to start the controller application using a

configuration file that contains at least one server definition.

An additional problem may be seen if the Start Server command issued

without any other options. This command will fail with the parsing error:

$ java -jar RDB$JDBC_HOME:RDBTHINCONTROL.JAR

rdbthincontrol> start server

Starting server ...

SQLException: Error parsing URL : //localhost:0/:S1000


53

5.8.2 Executor Initialization Problem



Depending on the system load it is possible that a newly created executor process

may not have fully completed its initial response to the Multi-process server before

the server checks the executor response data.

The incomplete response information may cause a problem with the initialization of

the connection that is waiting on the executor process creation.

The server log may show an RdbException logged for the thread where a zero value

for the executor version and/or instance may be seen similar to :

oracle.rdb.jdbc.common.RdbException: Executor images version

mismatch, server shr = 73202:20130501 , executor = 0:20130501

or :

oracle.rdb.jdbc.common.RdbException: Executor images version

mismatch, server shr = 73202:20130501 , executor = 0:0

It is possible, depending on the loading of the server system, that this problem may

in turn cause various access violations to be raised during the connection request,

which in turn will cause connection requests to be terminated in error.

5.8.3 Idle Client Termination breaks Subsequent Connection



A problem in the handling of idle client termination may cause subsequent

connection requests to fail.

During the termination of an idle client, the Multi-process server will force a

disconnection of the connection and terminate the executor process used by that

connection. This executor is no longer available for re-use, however a problem in

the executor cleanup code caused the server to incorrectly mark the executor as valid

for re-use. A subsequent connection request may get this unusable executor which

will cause the request to fail with an exception similar to following being raised in

the client application:

oracle.rdb.jdbc.common.RdbException: Connection lost :

java.io.EOFException

54

@rdb.Client.CONNECT_SECURE_V73

It is also possible that this same problem may eventually cause the Multi-process

server to hang requiring a forced termination of the server.

This problem only occurs when using Multi-process servers that have a non-zero

client idle timeout value set for the “cli.idleTimeout” server configuration option.

5.8.4 Controller STOPSERVER command Problem



A problem introduced in the first non-patched 7.3.2.0 release of JDBC may cause the

controller command line command STOPSERVER to raise an exception.

For example:

java -jar rdb$jdbc_home:rdbthincontrol.jar -stopserver 1818

raises the following exception:

SQLException: Error parsing URL : //localhost:0/:S1000


5.8.5 Hang on Connection to Server when using SSL



A code change made for the 7.3 release of JDBC may cause a client connection

request to wait indefinitely for a new connection when SSL is used.

If the client requests a normal , non-SSL connection to an SSL enabled JDBC server

or attempts to connect using SSL to a non-SSL JDBC server, the client connection

will wait indefinitely.

55

Prior to release 7.3 JDBC, the server would determine that there is a mismatch in the

SSL handshake after about 1 second , and force the connection request to terminate

immediately. A code change made during the 7.3 release accidentally disabled this

handling.

A workaround is to set the servers cli.idletimeout configuration attribute to a

positive value. If the SSL handshake does not complete within that time, the server

will abort the connection request. However, this also means that bound clients will

be forcibly disconnected if they are idle for longer than the cli.idletimeout

period.


5.8.6 DatabaseMetadata Pattern Matches and Nulls



In prior versions, Oracle JDBC for Rdb drivers did not correctly handle null values

for pattern parameters used in various DatabaseMetadata methods.

The JDBC standard states: "If a search pattern argument is set to null, that

argument's criterion will be dropped from the search".

The DatabaseMetadata methods that were not correctly handling nulls include:

GetProcedures()

GetProcedureColumns()

GetDomains()

GetFunctions()

These methods now handle null patterns correctly.

5.8.7 DatabaseMetadata Missing Index Information



In prior versions of Orace JDBC for Rdb, the DatabaseMetadata.GetIndexInfo()

method may fail to return information about existing indexes.

56

This method is used by tools such as the Rdb Addin for SQL Developer, to return a

list of indexes associated with each table found within the database schema.

Due to this problem, tools such as SQLDeveloper may fail to display existing

indexes.


5.8.8 Forced Client Termination may Crash Thin Server


It is possible that the forced termination of client applications by stopping or

destroying the client application process may cause an access violation to occur

within the JDBC server that the client was connected to.

Forcible termination of application using the thin driver may prevent the proper

rundown of the client side of the JDBC connection. The server will determine that

the client connection has been lost and will try to close down the connection in a

controlled manner, however it is possible that this cleanup processing may get

interupted by other server activities and may result in an access violation within the

server’s shared image.

The JDBC servers have now been changed to carry out more aggressive locking of

connection closedown operations which will prevent other server threads from

interfering with memory cleanup.

Oracle recommends to not forcibly close down client applications by process

termination using STOP/ID etc. Clients using the thin driver should either be

allowed to close-down normally or you may use the JDBC controller to stop the

client’s connection within the server, which will carry out a more regulated cleanup

of the connection.

5.8.9 Show Clients may cause NullPointerException in Server


Using the controller command SHOW CLIENTS may cause an exception to be

raised in the server the command was issued to:


oracle.rdb.jdbc.srv.ActionHandler.doControlPacket

oracle.rdb.jdbc.srv.SrvActionHandler.doControlPacket

57

This problem occurs infrequently and only happens when a client happens to

disconnect from the server within a small time window during the building of the

show client results by the server.

This problem does not affect the execution or viability of the active server, but will

cause the show client request to fail within the controller.

Note: As the ORCM application uses a variant of the thincontrol show clients

command it is also liable to show this same problem.


5.8.10 AccessViolation in MP Server


A problem in thread synchronization during some server control operations may

cause the MP Server to throw an Access Violation exception and terminate.

Under relatively rare circumstances the MP Server may fail due to an

AccessViolation while it is trying to deliver client or executor information to the

thincontroller or ORCM.

The following exception may be seen in the server log prior to the server

terminating:


address=00000000xxxxxxx, PC=000000000xxxxx, PS=0000xxxx

RDBJDBCMPSHR73 CJDBCCTX_M getDaMemUsage

RDBJDBCMPSHR73 CJDBCCTX_M getStmtMemUsage

This problem may be seen more often in MP Servers that create and destroy executor

processes frequently, and may happen immediately following the execution of

SHOW CLIENTS or SHOW EXECUTORS in a thincontroller session.

As ORCM also issues similar commands to obtain information about current server

use, it may also instigate this server problem.

The possibility of the server failure may be reduced by providing adequate free

executor processes for MP server clients to use, thus reducing the need for the server

to create or destroy executor processes.

58

5.8.11 Syntax Errors in Insert and Update Statements


A problem in the dbkey retrieval code generated by the JDBC drivers in order to

return the dbkey of newly updated or inserted record results in a syntax error.

The Rdb JDBC drivers add additional code to INSERT and UPDATE statements

during SQL statement preperation to return the dbkey of the resultant record. To

determine the position of the additional sql syntax, the JDBC drivers do a search for

any existing “INTO” clause. Unfortunately the drivers may mistake literals or

delimited identifiers may contain the whitespace-seperated word “INTO” for the

searched keyword resulting in incorrect SQL syntax being generated.

For example, the following SQL text may fail:

insert into tab9(fld1,fl2) values ('55555',' TEST INTO ')

The SYNTAX_ERR exception raised shows the modified SQL statement text,

%SQL-F-SYNTAX_ERR, Syntax error : in "insert into tab9(fld1,fld2) values

('55555',' TEST , DBKEY INTO '), ?"

A work around is to add the undocumented prefix “{nodbk}” at the start of the sql

text , which will disable the creation of the extra dbkey retrieval code:

{nodbk}insert into tab9(fld1,fl2) values ('55555',' TEST INTO ')


5.8.12 Statement.getGeneratedKeys() throws RdbException


Calling the Statement.getGeneratedKeys() method on an executed

statement that does not auto-generate keys should return an empty ResultSet.

The Rdb JDBC drivers do return an empty ResultSet, but incorrectly mark the

ResultSet as closed. Accessing this ResultSet may raise the following exception:

oracle.rdb.jdbc.common.RdbException: Closed Resultset

59

The Rdb JDBC drivers now mark the ResultSet as open.

5.8.13 SQLException thrown in ResultSet.isBeforeFirst()


When using the method ResultSet.isBeforeFirst() on a ResultSet produced by some

DatabaseMetaData methods and the Statement.getGeneratedKeys method, the

following exceptions may be raised:

java.sql.SQLException, Missing Curs in rdbFetchRow

or java.lang.NullPointerException


5.8.14 PreparedStatement executeBatch() Problems


A problem in the handling of batch entries for PreparedStatement may produce

incorrect results after batch execution.

Certain batch entries may be executed twice when

PreparedStatement.executeBatch() method is called. If the batch entry contains an

update or delete SQL statement, the entry may be executed twice in immediate

succession.

This may cause problems as the second instance of the execution may change the

intended behaviour of the original statement. For example the following SQL text :

update tab1 set F1 = F1 +1

will result in the F1 column being updated twice and end up as F1 +2.

In addition the updated count for that batch entry may be returned incorrectly, for

example the following SQL text:

delete from t1b where F1 = ?

60

will result in a value of 0 being returned in its update count value, as the first

execution of the statement will remove the records and the subsequent execution will

then return the value 0 as no records will be affected by its execution.

This problem only affects PreparedStatements that have update or delete SQL

statements batched. Normal Statements using executeBatch() are not affected.

Another problem may be see when executing an update SQL statement using the

PreparedStatement executeBatch. If the statement is not a singleton operation, that

is, if it updates more than one record, the batched statement may fail with the

following error:

java.sql.BatchUpdateException: Problem with one or more batched query

And show the following underlying Rdb error:

%RDB-E-MULTIPLE_MATCH, record selection criteria should identify only one

record; more than one record found

These problem have now been fixed.

▲contents



5.9.1 Server Options List Inheritance Release 7.3.2.0.0.

When a XML-Formatted configuration file is used, many server characteristics may

be inherited from either the DEFAULT or the DEFAULTSSL server definitions.

If a characteristic is specified within a specific server definition, that specification

will take precedence. If the characteristic is not specified in the specific server

definition then if it exists in the default server definition it will be inherited.

In previous versions of Oracle JDBC for Rdb, some characteristics were not

inherited and even if specified within the default server definition they would not be

passed on to other servers in the same configuration file.

61

The characteristic that did not get inherited were all list options that included:

Server Configuration List Options

Option

<allowDatabase name=”database-name”>

<allowPrivUser name=”user-name”>

<allowUser name=”user-name”>

<deny sql =”sql-pattern”>

<enableEvent name=”event-name”>

List options provide a mechanism to specify one or more named options that may be

used by the server for validity, event checking and other operations. Each server

definition may contain multiple values for the same named option, for example,

multiple allowUser entries can be used to provide a list of users that will be

allowed to access that server.

Starting with release 7.3.2.0, the list options specified above may now be inherited

from the default server.

The inheritance of these options is cumulative, in that the resultant list of options is

the combination of those present in both the default and the specific server

definitions.

If a server definition already has an option with the same type and name, it will not be

replaced.

Note

An error in prior documentation specified that the enableEvent

configuration option would be inherited; this is not true for versions of

Oracle JDBC for Rdb prior to release 7.3.2.0.

5.9.2 Oracle JDBC for Rdb Manager Server

Release 7.3.2.0.0.

The Oracle JDBC for Rdb manager server is a server-side component that services

JDBC management requests.

The main purpose of the manager server is to provide a mechanism that will allow

JDBC servers to be started up on nodes remote to the one the controlling application,

such as the Oracle JDBC for Rdb controller, is running.

62

Please see the Oracle JDBC for Rdb Manager Server section of the Oracle JDBC for

Rdb User Guide for more information on manager servers.

5.9.3 Restricting Server Access by IP

Release 7.3.2.0.0.

JDBC servers may now check incoming connection IPs to determine if the client

connection will be accepted.

New server configuration options:

● <allow ip = “<valid ip mask>” />

● <deny ip = “<valid ip mask>” />

Using the new server configuration options you may restrict access to the server by

having the server check the originating IP of the client connection to either accept or

reject that IP. The valid IP mask may be either a simple IP specification or may be

JAVA regular expression.

See the Restricting Server, Database and Operational Access section of the Oracle

JDBC for Rdb User Guide for more details.

5.9.4 Executor Balancing

Release 7.3.2.0.0.

By default, the multi-process server uses a first-in/first-out (FIFO) scan of its free

executor list to select the next executor process to allocate to the new connection

request.

Oracle JDBC for Rdb now allows you to choose from several new executor

balancing protocols.

Please see the Executor Balancing section of the Oracle JDBC for Rdb User Guide

for more information.

5.9.5 MinFreeExecutors

Release 7.3.2.0.0.

Multi-process servers may now be configured to maintain a minimum set of free

executor processes when the server is cleaning up executor processes that have

exceeded their maximum idle time.

63

Please see the Multi-process Server Configuration Options and the Executor

Maintenance sections of the Oracle JDBC for Rdb User Guide for more information.

5.9.6 Executor Reuse

Release 7.3.2.0.0.

Multi-process servers may now be configured to change the way they reuse executor

processes.

Please see the Multi-process Server Configuration Options and the Executor Reuse

sections of the Oracle JDBC for Rdb User Guide for more information.

5.9.7 LogFile Patterns

Release 7.3.2.0.0.

The server logfile configuration option provides an absolute or relative file

specification to use as a log file. This specification may also be a special pattern, in

which case certain key character sequences may be used in conjunction with normal

ASCII filename characters to specify the automatic creation of the log file name.

See the Logfile Pattern section of the Oracle JDBC for Rdb User Guide for more

information.

5.9.8 New Server Configuration Option retainRdbSQLState

Release 7.3.2.0.0.

In keeping with third party application requirements, during the processing of

INSERT statements, the SQL State returned when using the

SQLException.getSQLState() method is set to the value “S1000” indicating

an error has occurred. The message text of the exception contains the details of the

error raised.

Oracle Rdb delivers more specific SQL State values including Rdb specific

(“Rxxxx”) state values, but these may be suppressed by JDBC.

The server retainRdbSQLState configuration option may be used to change this

behaviour. When set to “true”, JDBC will return the original SQL State raised by the

underlying database.

For example:

64

<server

name = "DEFAULT"

type = "RdbThinSrv"


retainRdbSQLState = "true" />

See the Server Configuration Options section of the Oracle JDBC for Rdb User

Guide for more information.

5.9.9 New Connection Option app

Release 7.3.2.0.0.

To help identify applications usage of JDBC servers, a new connection option, app,

has been introduced.

For example:


"jdbc:rdbThin://bravo:1755/my_db_dir:pers@app=MyOwnApplication",

user, pass);


more information.

▲contents



5.10.1 Event Flag Problem with long Executor name Prefixes


During the creation of an executor process by the Multi-process server, the server

attempts to provide a unique process name for the executor so that it may be easier to

recognize when using system tools such as SHOW SYSTEM.

65

By default, the server will create the process name by utilizing the server name and

the unique connection id.

To allow better control of process naming, the prefix of the executor process name

may also be explicitly specified by using the server configuration option

srv.execPrefix. The server will use this prefix and append a sequence number

to it in an attempt to provide an unique name.

As there is a limit of 15 characters allowed for OpenVMS process names it is

possible that longer prefixes may cause the unique portion of the name to cycle

around faster and increase the probability that a process name may be chosen that

matches an existing running executor process.

If the process name matches an existing process, the startup of the new executor will

fail and the executor process will not be present to acknowledge startup completion

to the server.

Normally an executor process failure will be correctly caught by the server and the

appropriate exceptions raise, however if the failure happens after the process is

correctly created but before this initial acknowledgement, the server will not see the

failure for several minutes.

Eventually, an exception similar to the following will be raised:

oracle.rdb.jdbc.common.RdbException: Executor not responding on

EFN 66 EFNMASK 4 after 302 seconds

This message will be written to the server log. In addition other messages describing

the reason for the failure may be found in the log files associated with the executor in

the JDBC log area which by default is pointed to by the logical name

RDB$JDBC_LOGS.

Multi-process servers have now been changed to try to better handle process name

clashes by doing further name checking on the running system prior to attempting to

start the new process.

5.10.2 Executor Process Termination Problem


During the normal shutdown of an executor, the Multi-process server will request the

executor process to self-terminate to ensure that all resources are correctly released

prior to the OpenVMS process terminating.

After this self-termination is requested, the server will then check to see if the

process is still there, and if it is, will forcibly stop the process.

66

In heavily utilized systems it is possible that the executor process self-termination

has not completed by the time the server does its secondary check and subsequent

process termination. In this situation the process may get into a “limbo” state where

it fails to closedown correctly and depending on when the second termination is

issued the executor process may get trapped in an endless exception handler loop.

The result of this problem is that the executor process may start to consume large

amounts of CPU, or may remain idle on the system without releasing resources. In

addition this process may cause other problems if the server is restarted and tries to

use the same process name for a newly created executor.

This has now been fixed, the server will now allow a greater amount of time for the

self-termination to occur before forcing an image exit.

5.10.3 Some Server Characteristics not Correctly Inherited from DEFAULT

BUG 14510287.


Some server characteristics are not correctly copied from the DEFAULT or

DEFAULTSSL server definitions when named servers are used.

For example if mycfg.xml contains the following:

.

.

.

<server

name="DEFAULT"

type="RdbThinSrv"

url="//localhost:1701/"

maxClients="-1"


>

<allowUser name = "JDBC_USER"/>

</server>

<server name="srv1" />

.

.

.

Starting the server using this configuration and using the –verify switch to verify the

server characteristics:

67

java –jar rdb$jdbc_home:rdbthinsrv.jar –cfg mycfg.xml –name "srv1" -verify

will raise the following exception:


at oracle.rdb.jdbc.srv.RdbSrv.logServerDetails

Although the restrictAccess option is correctly inherited , the allowUser

option from the DEFAULT server is not, causing this problem.

Other server characteristics may also fail to be inherited, although several may not

cause immediately obvious problems.

The following is a list of server configuration options that are not correctly inherited

from either of the default server definitions:

Server Configuration Option

Option

<allowDatabase name=”database-name”>

<allowPrivUser name=”user-name”>

<allowUser name=”user-name”>

<deny sql =”sql-pattern”>

<enableEvent name=”event-name”>

A workaround is to place these options explicitly in each server definition that

requires them.


5.10.4 Memory Problem with Pool Servers and Java 1.6.0-2


A problem in Java 1.6.0-2 in the recovery of memory when byte streams are garbage

collected may prevent the associated memory from being recovered correctly which

may lead to a memory leak in the Java VM. This problem does not occur if the byte

streams are explicitly closed prior to garbage collection.

JDBC servers try to ensure that all byte streams used are correctly closed prior to

releasing the socket connection to the client, however, a problem within Pool Servers

may cause the explicit closure of a byte stream to be bypassed.

68

The failure of the byte stream to be garbage collected correctly may eventually result

in Java VM problems and may cause an exception similar to the following which

will terminate the Pool Server:

%SYSTEM-F-ASTFLT, AST fault, SP=0082D250, param=7FFEFFC8,

.

.

.

The Pool Server has now been modified to ensure that all byte streams are correctly

closed prior to socket closure.

OpenVMS Java engineering is aware of this problem.

5.10.5 Exception not Thrown when Record Locked during Update


Normally, if JDBC finds a record locked during an insert, update or delete Statement

operation, a Record Locked exception will be raised.

However, a problem in the handling of record locks by the Native driver prevents the

driver from raising a locked record exception correctly.

Instead, if a locked record is found, the update operation will be aborted and the

statement’s update count, obtained using the

Statement.getUpdateCount(), method will be set to -4.

Although an exception is not raised, you may still determine if a record lock has

occurred by calling the Statement.getUpdateCount() after calling the

statement execute. If the value returned is -4, a locked record was found. If no

locked records were found, the value returned will be the number of records affected

by the update.

This problem affects the Native driver only. The Thin driver is not affected.

This problem has been fixed.

5.10.6 Classpath Documentation Error


69

In prior versions of Oracle JDBC for Rdb, due to a typographic error, the Driver

Class section of the User Guide referenced the classpath for the JDBC drivers

as:

Driver Class Classpath

oracle.jdbc.rdb.rdbThin.Driver or

oracle.jdbc.rdb.rdbNative.Driver

The was incorrect, and should read:

Driver Class

Classpath oracle.rdb.jdbc.rdbThin.Driver

or oracle.rdb.jdbc.rdbNative.Driver

The subsequent code example in the same section shows the classpath correctly.

A similar error may be found in the following User Guide sections:

Results Class

Blob Class


▲contents



5.11.1 “Owner” may be used in XML Configuration Files Release 7.3.1.0.0.

70

The keyword OWNER may be used in certain sections that require a username in the

XML Server configuration file.

The keyword OWNER specifies the OpenVMS account username of the account

used to start the server, and may be used instead of a username in sections such as

AllowUser or AllowPrivUser. For example:

.

.

. <server

name = "rdbthnsrv8"

type = "RdbThinSrv"



srv.showPoll = "true"

allowAccessToCL = "true"

>

<allowUser name = "jdbc_user"/>

<allowUser name = "owner"/>

<allowPrivUser name = "owner"/>

</server>

.

.

.

5.11.2 SQL Statement Restriction and Denial

Release 7.3.1.0.0.

Servers may be configured to restrict or deny certain SQL statements.

See section Restricting SQL Statements in the Oracle JDBC for Rdb User Guide for

more details.

5.11.3 Event Notification Release 7.3.1.0.0.

Servers may be configured to notify enlisted event listeners of events as they occur

on the server. Events such as connect and disconnect and memory usage may be

monitored.

See section Event Logging and Notification in the Oracle JDBC for Rdb User Guide

for more details.


71


5.12.1 Multi-process Server Connection Hang when Executor Dies Fixed in Instance Build 20111110 release 7.3.1.0.0.

If an executor process dies or is forcibly terminated it is possible that the associated

connection within the Multi-process (MP) server may hang indefinitely waiting on a

response from the non-existant executor.

Executor process termination is notified to the server through several different

internal mechanisms, however inherent in Pthread mutex operations is the possibility

that locking operations may prevent the associated thread from acting on the

termination notification. If this termination notification is missed or blocked, the

thread may wait indefinitely.

The MP server has been changed to include an executor process sweeper thread that

will check each active connection utilizing an MP executor to determine if it is still

viable.

If the sweeper find that an executor has been flagged as terminated but the connection

is still active and has not been rundown, the thread for the active connection will be

forcibly terminated causing a socket exception on the client side.

The new server parameter srv.lostExecSweep may be used to specify the

frequency of the sweeper thread execution.

See your Oracle JDBC for Rdb User Guide for more information on the server

parameter srv.lostExecSweep.

5.12.2 Multi-process Executor Process Terminates Unexpectedly


A problem in the server viability checking carried out by the executor process may

cause the executor process to self-terminate unexpectedly.

If an executor is associated with an active connection and has remained idle for a

predefined amount of time as detailed below, it may incorrectly determine that the

parent server is no longer running and self-terminate.

The amount of time the executor will remain in an idle state before termination

depends on the MP server attribute srv.MPMaxTries.

If the executor is is idle for more than srv.MPMaxTries * 3 seconds, it may

self-terminate, thus if the MP server is using the default srv.MPMaxTries of 500

72

then the executor will self-terminate after 25 minutes of idle time if it still holds an

active connection.

This problem only effect executors associated with active connections and not

executors within the MP server free executor pool.

A work-around is to set the srv.MPMaxTries to a very large number to delay

this termination.

This problem was introduced in release 7.3.0.0 and is now fixed.

5.12.3 Pooled Server AutoRestart Problem


A problem introduced in release 7.3.0.2.0 (V7.3-02) may cause the Pool server to

incorrectly flag its children pooled servers as unavailable and prevent new

connections from being made to these servers.

If a pooled server has “Autorestart” enabled, the Pool server will periodically check

to see if the server is viable and if it determines that the server is no longer

responding, the Pool server will mark the pooled server as unavailable and then

attempt to restart it.

A problem in how the Pool server determines the viability of the child server causes

the Pool server to incorrectly mark the child as unavailable. The attempted restart

will fail as the original pooled server is still present and connected to the socket that

the new instance is trying to allocate. Thus the pooled server will then remain

marked as unavailable and will no longer participate in the pool.

This problem will prevent all pooled servers marked as “Autorestart” from

participating in the server pool.

A work-around is to set Autorestart=”FALSE” for each of the pooled servers

in the server XML configuration file. This will prevent the Pool server from

checking and incorrectly marking as invalid.

5.12.4 getGeneratedKeys() and OutOfBounds Exception


The call to ResultSet.getNext() on a resultset returned by the

Statement.getGeneratedKeys() may throw the following exception:

java.lang.ArrayIndexOutOfBoundsException: -1

73

This exception only occurs when the getNext() method is called on the resultset

after a prior getNext() returned false indicating an end of recordstream.

The following code example will show this problem:

.

.

.

ps2.execute();

rs = ps2.getGeneratedKeys();

if (rs == null)

log ( "No resultset from execute");

else

{

while (rs.next())

{

.

.

.

}

rs.next(); // this will throw an array exception

}

.

.

.


5.12.5 Multi-process Server fails when Persona Used


A problem introduced in 7.3.0.2.0 (V7.3-02) may cause the Multi-process server to

terminate while attempting to start an executor process.

This problem only occurs if the server has the Persona attribute set.

If the Persona attribute is designated for the server, executor processes will be

created using the correct Persona however a problem in determining the process

identifier of the executor process may cause a Access Violation within the server

process which will terminate the Server image.

The following is an example of the type of exception raised:

#

# An unexpected error has been detected by HotSpot Virtual Machine:

#

74

# %SYSTEM-F-ACCVIO (0xc) at pc=35C5481, pid=564797798, tid=2070302400

#

# Java VM: Java HotSpot(TM) Server VM (1.5.0.20 08/30/2010-13:43 IA64 mixed

mode)

# Problematic frame:

# C [RDBJDBCMPSHR73+0xffffffb2]

#

# An error report file with more information is saved as

hs_err_pid564797798.log


address=0000000000000000, PC=000000000113DC30, PS=0000001B

%TRACE-F-TRACEBACK, symbolic stack dump follows

image module routine line rel PC abs PC

JAVA$HOTSPOT_SHR OS print_hex_dump 133224 0000000000001EF0

000000000113DC30

.

.

.

----- Above condition handler called with exception 0000000C


address=0000000000000010, PC=00000000035C5481, PS=0000001B

----- End of Exception message

0 FFFFFFFF80393F42

FFFFFFFF80393F42

RDBJDBCMPSHR73 RJDBC_PROC rjdbc_proc_GetPid

42606 00000000000029A1

00000000035C5481

RDBJDBCMPSHR73 RDBJDBC Java_rdb_JNI_StartProcess

The only work-around is to not use Persona for Multi-process servers.


5.12.6 RestrictAccess being Applied even when Disabled


The JDBC user quide specifies that if the server attribute restrictAccess is not

defined or is set to false then database and user restrictions will not be carried out.

A problem in how server restriction policy was applied may cause restrictions to be

applied even if the server attribute restrictAccess is not defined or is set to false.

If there is at least one allowDatabase element in the server specification then

database restrictions were applied to the server. Similarly, the presence of

allowUser entries would cause the appropriate restrictions to be applied

irrespective of the restrictAccess settings.


75

5.12.7 Multi-process Problem in the Native Driver


Due to changes in thread interleaving introduced in Java JDK6.0 it is possible that

applications using the multi-process option of the rdbNative driver connections

may no longer work correctly when run in a JDK6.0 environment

Interference may occur between simultaneously running threads accessing JDBC

connections within the same application object instance which may cause Access

Violations within the application.

The nature of the problem depend very much on where the interference occurs and

what each thread was trying to do at the time, however more generally the problems

may be seen during connection startup.

Various Access Violations, problems with OpenVMS Event Flags and Pthread

mutex problems have been noted during connection, with the subsequent prevention

of JDBC connections being made correctly or resulting in the hanging or the

termination of the running application.

The JDBC rdbNative driver has now been changed to carry out more aggressive

thread locking at the initial handshake phase of the JDBC connection with the Multi-

process executor, to prevent these inter-thread problems.

This problem affects the rdbNative driver only, applications using the rdbThin driver

connecting to Multi-process servers are not affected.

▲contents



5.13.1 Server Network Keep Alive

Release 7.3.0.2.0.

The ability to set SO_KEEPALIVE on the socket used on client–side to communicate

with a server has been available for a number of releases. This release introduces the

same feature on the server-side socket used to communicate back to the client.

See srv.networkKeepAlive in the Server Configuration Options section of

your Oracle JDBC for Rdb User Guide for more information.

76



5.14.1 Null pointer Exception during Server viability check by Pool Server


When the Pool server checks to see if a pooled server is available and has a free client

slot, it is possible that a problem in a disconnect code feature that was added for the

Oracle JDBC for Rdb 7.3.0.0 (V7.3) upgrade may cause a

NullPointerException to be raised by the pooled server.

The pooled server recovers from this exception, but an entry showing the exception is

written to the server log.


5.14.2 MINUS and INTERSECT Problem


The use of the set operators MINUS and INTERSECT may cause the JDBC SQL

parser to incorrectly allocate dbkey and alias variables within the compound

statement which may raise SQL syntex errors when the statement is compiled by Rdb.

For example, an SQL syntax error will be raised by the following simple query:

select * from employees MINUS select * from employees;

SQLException: in <rdbjdbcsrv:prepare_stmt>

F-RELNOTDEF, Table MINUS is not defined in database or schema:42000


5.14.3 Memory Leak when Executor Process fails to Run


77

A small memory leak has been found in the Multi-process(MP) Server when the

server attempts to start a new executor process but fails due to a System or a resource

problem. An example of a resource limitation may be insufficient process slots for

the additional executor process.

The MP Server correctly signals the exception however it then fails to recover a small

amount of memory (157 bytes) which may cause process memory depletion if this

situation is repeated many times within the life-time of the server.


5.14.4 Lost Executor not caught by Multi-process Server


If an executor process terminates unexpectedly while still connected to an active

connection, its is possible that the Multi-process server parent process may not get

notification of the terminations and will wait indefinitely for the executor process to

respond.


5.14.5 Multi-process Native Driver Problem


During the creation of connection contextes when using the Multi-process feature of

the Native driver , the driver did not adequately synchronize the connect operation.

This may result in unexpected and incorrect behaviour during connection if multiple

threads are concurrently trying to connect to databases within the same process.


5.14.6 PreparedStatement.getGeneratedKeys() Problem


On the second or subsequent execution of a PreparedStatement, the call to the method

getGeneratedKeys() will fail with the following exception:

oracle.rdb.jdbc.common.RdbException: Closed Resultset


78

5.14.7 Excessive IOs during Metadata Retrieval


Several SQL queries used by JDBC to retrieve metadata associated with compiled

queries may cause excessive IOs on database systems where the column or table

definitions are volatile and have had many revision during the life of the database.

This problem was introduced in 7.3.0.2.0 and has now been fixed.

5.14.8 Syntax error SQL-F-CONVARUND


A problem in the determination of the correct table or alias name to use for dbkey

references in the additional SQL syntax added by JDBC to allow correct record

identification may produce SQL syntax exceptions to be raised.

This problem prevents the query from executing correctly, for example:

select * from employees e left join degrees on

(degrees.employee_id = e.employee_id)

will raise the following error:

SQLException: SQLState(RR000) vendor code(-1)

@ oracle.rdb.jdbc.common.RdbException: in <rdbjdbcsrv:prepare_stmt>

@ %SQL-F-CONVARUND, Column qualifier EMPLOYEES is not defined

This problem only occurs in queries that reference two or more tables.


5.14.9 Multi-process Server Hang


The Multi-process (MP) server may hang during Resultset close operations.

79

A problem introduced in 7.3.0.2.0 may cause the server to try to execute code outside

the standard application space and cause either a hang in the server process or an

Access Violation.

The MP server may encounter this problem at any time while trying to close

ResultSets , however in the majority of cases the ResultSet close will operate

correctly without problem.

The problem may be seen more readilly in very active MP server processes.

Spurious Rdb Dispatcher errors have also been noted due to this problem.


5.14.10 SSL Socket Intrusion Problem


Data injection into an existing open SSL sockets may cause problems with SSL-

enabled Thin Servers.

Socket and Port security scanning software may inject data into exiting open SSL

socket channels that may cause SSLExceptions on that socket. Depending on the

state of the Thin Server at the time of the injection, this intrusion may be handled

incorrectly causing the server process to use excessive CPU.

The SSL connection is not compromised by the intrusion attempt, and the underlying

database connection is not affected.


▲contents



5.15.1 Reopen Server Log files using Poll Subcommand

Release 7.3.0.1.0.

80

The thincontroller now has the capability to request servers to re-open their log files

allowing the previous version of the log file to be examined or copied.

Due to a restriction within Java on OpenVMS, log files opened by a Oracle JDBC for

Rdb server cannot be be read or copied while the log files are currently being used by

the server.

The thincontroller from Oracle JDBC for Rdb 7.3.0.1.0 (V7.3-01) , or later from later

releases, used in conjunction with Oracle JDBC for Rdb servers from 7.3.0.1 and later

releases will allow a control user to request that the server re-open its log file by using

the POLL REOPENLOGS subcommand.

See the POLL Sub-commands section of the Oracle JDBC for Rdb User Guide for

more information.

▲contents


This section describes software errors corrected in Oracle JDBC for Rdb release

7.3.0.1.x.

5.16.1 Show Clients not Showing Column Names


The Show Clients command within the Controller displays information about clients

currently using JDBC servers.

A problem introduced in Oracle JDBC for Rdb 7.3.0.0.0 prevents the column header

information from being displayed correctly.

The data for each client is still displayed correctly but without column identification.

rdbthincontrol> show active clients

Server rdbthnsrv3 (//192.168.1.3:1706/)

00000004*

<CONTROL CONNECTION>

jdbc_user

192.168.1.3:1459

0x16A0(5792)

0x16A0(5792)

81

2010-03-15 09:59:59.296 : INIT_CONTROL

0 00:00

0

1970-01-01 10:00:00.0

rdbthincontrol>

The following shows how the display should look:

rdbthincontrol> show active clients

Server rdbthnsrv3 (//192.168.1.3:1706/)

RDB$CLIENT_ID : 00000006*

RDB$URL : <CONTROL CONNECTION>

RDB$USER : jdbc_user

RDB$IP : 192.168.1.3:1665

RDB$PID : 0x16A0(5792)

RDB$PID_AT_EXECUTOR : 0x16A0(5792)

RDB$LAST_SQL :

RDB$LAST_ACTION : 2010-03-15 10:55:01.312 : INIT_CONTROL

RDB$LAST_EXCEPTION :

RDB$TIME_SINCE_LAST_ACTION : 0 00:00

RDB$MINUTES_SINCE_LAST_ACTION : 0

RDB$LAST_OPEN : 1970-01-01 10:00:00.0

rdbthincontrol>

5.16.2 Batched Statement Fails with MULTIPLE_RECORDS Exception


To provide additional JDBC driver functionality in Oracle JDBC for Rdb 7.3.0.0.0,

during the preliminary parsing of SQL update or insert statements, the Oracle JDBC

for Rdb drivers may append extra SQL syntax to the query to deliver the dbkey of the

updated or inserted record back to the driver for later processing.

If during the execution of that query, Rdb raises a MULTIPLE_RECORDS

exception, the JDBC drivers will re-issue the statement again without the dbkey

retrieval clause, allowing the update statement to complete succesfully.

If however, the update statement is issued as a batched statement by using the

Statement.addBatch() method, the additional dbkey retrieval code may cause

that batch statement to fail. In which case Oracle JDBC for Rdb does not re-issue the

query and the update will not be done.

The Rdb JDBC drivers have been changed to not add the extra dbkey retrieval code if

the statement is being processed as a batched statement.

82

5.16.3 Incompatibility between 7.3.0.0 Thin Driver and Prior Release Servers


Ideally, when upgrading Oracle JDBC for Rdb, all components on both the server

systems and all client systems should be updated to the same version at the same

time.

Although not documented nor supported as a feature, Oracle JDBC for Rdb attempts

to maintain upward and downward compatibility between its servers and the Rdb

Thin drivers.

However a change in the initial connection hand-shake between the Thin driver and

server introduced in 7.3.0.0 (V7.3) may prevent applications using the 7.3.0.0.0 Thin

driver from connecting correctly to Oracle JDBC for Rdb servers from earlier Oracle

JDBC for Rdb versions.

The connection of applications using Thin drivers from release prior to 7.3.0.0.0 to a

7.3.0.0.0 server should still work correctly.

These incompatibility issues have now been resolved.

5.16.4 Global Memory leak when Executors are Run-down


Bug 9787051

A problem in the release of global shared memory when an executor process is run-

down by a Multi-process (MP) server may cause the server to eventually run out of

global memory.

If this happens the following exception will be raised:

System Error : Insufficient global memory

When a MP server runs out of global memory it cannot start-up new executor

processes.

As this problem occurs when an executor is run-down when it is no longer required, a

work-around for this problem is to increase the maxFreeExecutor count for the

server to a number that will be sufficient to handle the peak load of the server and to

increase the server sharedMem appropriately.

83


5.16.5 Autorestart on Pooled Servers not Working

BUG 9656632


When a Pool server starts a child server that is marked for autorestart, the child server

information is placed on a queue that is periodically checked by the Pool server to

ensure the child server is still viable.

If the Pool server finds that the child is not available and that child server is marked

for autorestart, the Pool server will try to restart that server.

A problem was introduced in Oracle JDBC for Rdb 7.3.0.0.0 (V7.3) within the Pool

server code that caused the Pool server to incorrectly mark a newly started child

server as “failed in startup”.

If the pool server thinks that the child server did not start correctly the server will not

be placed in the check queue used to determine if the child server is still available.

This in turn means that the pool server will not automatically restart the child server if

the child server process subsequently dies.

This has been fixed.

5.16.6 setFetchSize() hint Ignored by PreparedStatement


In prior versions of Oracle JDBC for Rdb, the execution of setFetchSize()

method on a PreparedStatement would be silently ignored.

The connected server process would determine the fetch size of subsequent

operations on the PreparedStatement and ignore the fetch size hint provided.

This behavior has now been changed. The fetch size hint provided by the execution

of the setFetchSize() method on a PreparedStatement prior to the

execution of the statement may now be taken into account by the server when

determining the number of records that should be sent in one network IO operation.

As this is only a hint, the connected server may still choose to ignore it.

84

5.16.7 Missing Stmt Exception on Subsequent Execution of PreparedStatement


The second or subsequent execution of a PreparedStatement may raise the


java.sql.SQLException: Missing Stmt in rdbSelect

@rdb.Client.SELECT

This exception may be raised on any subsequent executions of the same

PreparedStatement object if the prior execution of the same statement retrieved a

greater number of records than the connected server’s fetch size.

The following example will show this errant behavior if the fetchSize specified for

the connected server is less than 10 rows:

.

.

.

try

{

ps = conn.prepareStatement(

"select last_name from employees limit to 10 row");

ResultSet rs = ps.executeQuery();

while (rs.next())

{

System.out.println(" result1 <" + rs.getString(1) + ">");

}

rs = ps.executeQuery(); //<<<<<< this would fail

while (rs.next())

{

System.out.println(" result2 <" + rs.getString(1) + ">");

}

}

finally

{

if ( ps != null ) ps.close();

}

.

.

.

This problem was introduced in 7.3.0.0.0 (V7.3) JDBC and only occurs when using

the Oracle JDBC for Rdb thin driver.

5.16.8 EOFException on READ_ROW with Nested Statements

85


A problem in statement context handling may cause the following exception to be

raised:

java.io.EOFException @rdb.Client.READ_ROW

This problem may occur when the following conditions are met.

1. A select query is executed as the outer query

2. An update statement is executed as an inner statement

3. The fetch size of the select query is less than the total number of records

returned by the query

4. Fetch Size number of record have already been returned by the outer query

An example of code which may show this problem is shown below:

.

.

. Statement stmt = conn.createStatement(); stmt.setFetchSize(10); // less than the total number of records

PreparedStatement ps = conn.prepareStatement (

"update job_history set supervisor_id = '00245' where "+

"employee_id= ?");

ResultSet rset = stmt.executeQuery(

"select employee_id from employees limit to 20 rows");

int i = 0;

try

{

while (rset.next())

{

i = i + 1;

System.out.println(i + " " +

rset.getString("employee_id"));

ps.setString(1, rset.getString("employee_id"));

ps.executeUpdate();

}

}

catch (SQLException sqle)

{

sqle.printStackTrace();

}

.

.

.

5.16.9 DatabaseMetaData.getUDTs()


86

Although Rdb does not currently support UDTs you may still retrieve the column

descriptions from the empty Resultset produced when the method

DatabaseMetaData.getUDTs() is called.

However a problem in the retrieval of the last column from the ResultSetMetaData

may result in a Null Pointer exception.



5.16.10 Using Multi-process with the JDBC Native Driver


A problem was introduced in Oracle JDBC for Rdb 7.3.0.0.0 (V7.3) preventing the

use of the multi-process option within JDBC Native driver connections.

When a connection is attempted using the JDBC Native driver with a connection

URL that contains the @multi-process=true switch the following exception

was raised:

INFO: server version needs to be V7.3 or later


5.16.11 Prepared Statement not Closing underlying Cursor


A problem introduced in Oracle JDBC for Rdb 7.3.0.0.0 (V7.3) prevents

PreparedStatements containing select statements from correctly closing cursors when

the statement is re-executed but no results set read was issued on the previous

execution of the same prepared statement instance.

The following exception may be raised:

%SQL-F-CURALROPE, Cursor C_xxxxxxxxx was already open

The following code snippet show an example of the type of code that may show this

problem:

.

87

.

.

String sql = "select dbkey from employees";

ResultSet rs;

int maxi = 10;

PreparedStatement st=conn.prepareStatement(sql);

for(int j=0;j<maxi;j++)

{

rs=st.executeQuery();

rs.close(); // note: no read on the resultset

// issued within the iterating block

}

.

.

.

The problem would be seen on the second and subsequent iterations of the

executeQuery.


5.16.12 Release Statement Synchronization Problem


Oracle JDBC for Rdb release 7.3.0.0.0 (V7.3) introduced improvements in the

network handling of selection statements. A problem in the optimization of network

IOs introduce in release 7.3.0.0.0 may produce the following exception during

statement release:

oracle.rdb.jdbc.common.RdbException: received 503316549(0x4500001E)

: expected 1 @rdb.Client.RELEASE_STMT

This problem may occur only when the following conditions have been met:

1. A select statement has been executed.

2. Statement select optimization has not been turned off.

3. The total size of a FetchSize number of records is greater than 9600 bytes.

4. The statement is closed before any records have been read from the

associated resultset.


5.16.13 Unitialized SQLCA Block in MP server


88

A problem in how internal SQLCA blocks were initialized by the Multi-process

server may cause a nested exception to be raised during exception handling.

The SQLCA block failed to initialize correctly which may cause incorrect message

size information to be used when the server tries to dump out the underlying error

message, which in turn may cause the server to try to access incorrect memory

addresses.

This problem may lead to various access violation exceptions including :

%CXXL-F-TERMINATE, terminate() or unexpected() called

Improperly handled condition, image exit forced by last chance

handler.


5.16.14 Controller SHOW CLIENTS and MP Server Problem


Note

This problem was reported fixed in Oracle JDBC for Rdb release 7.3.0.0.0 but

unfortunately, the fix was incomplete.

When the command SHOW CLIENTS or SHOW ALL CLIENTS is issued within the

Controller, if the recipient server is a Multi-process server (MP Server), an executor

process may be started up by the server to execute the request.

A problem in the executor code prevents the executor from closing down correctly if

the maximum number of free executors has already been reached for the MP Server.

This new executor process will not be added to the free executor pool, but will not be

shutdown and thus will remain active on the system.

If the controller SHOW CLIENTS command is issued again, the number of executor

processes may increase until an Open VMS quota or process quota is exceeded.

This problem has now been fixed.

89

5.16.15 Documentation Error – Record Streaming

In the Oracle JDBC for Rdb 7.3.0.0.0 release notes the Performance Enhancements

subsection of the New Features section mentions that Record Streaming is available

from release 7.3.0.0.0 onwards.

Unfortunately a number of problems required the removal of this feature prior to the

7.3.0.0.0. release, but the 7.3.0.0.0 releases notes were not correctly updated to

remove reference to the feature prior to product shipment.

This feature may be made available in a later release of Oracle JDBC for Rdb.

5.16.16 ResultSet.updateRow() and ResultSet.deleteRow() Problem


Bug 9668574

A problem introduced in release 7.3.0.0.0 prevents ResultSet.updateRow() and

ResultSet.deleteRow() from working correctly.

During the execution of the update one of the following errors may be raised:

Internal Error

or

SQL-F-SYNTAX_ERR, SYNTAX ERROR


5.16.17 Problem using Column Renaming with dbkey

Fixed in Instance Build 20110127.

Changes to SQL statetment parsing introduced in release 7.3.0.0.0 inadvertantly

introduced a problem in queries that access the dbkey of records directly.

For example the following query:

90

select dbkey as ID, last_name from employees;

may cause the following exception to be raised:

Exception: Unhandled [email protected] :

java.lang.ArrayIndexOutOfBoundsException:S1000

This problem only occurs when the dbkey is retrieved as the first named column of

the selection expression and the column name is changed by using the ‘AS’ clause.

Some simple work-arounds are:

1. move the dbkey clause to the second or subsequent column reference:

select last_name, dbkey as ID from employees;

2. Remove the column renaming:

select dbkey, last_name from employees;


5.16.18 Class cast error on Scaled integer Retrieval after ResultSet.insertRow()


During the conversion of a scaled integer invoked by a ResultSet object retrieval the

following error may be raised:

oracle.rdb.jdbc.common.RdbException: Connection lost

: java.lang.Long

@rdb.Client.GET_INT_VAL

This problem occurs when:

● The object retrieved is a scaled integer

● Immediately prior to the retrieval a ResultSet.insertRow() was carried out on

the same ResultSet

The following code example will show this problem during the call to the getInt()

method:

91

.

.

.

Statement stmt = conn.createStatement();

ResultSet rset = stmt.executeQuery("select * from JOBS");

rset.moveToInsertRow();

rset.updateString("JOB_CODE", "JAVA");

rset.updateString("WAGE_CLASS","1");

rset.updateString("JOB_TITLE","Java Tester");

rset.updateInt("MINIMUM_SALARY",90000);

rset.updateInt("MAXIMUM_SALARY",250000);

rset.insertRow();

while (rset.next())

{

System.out.println(rset.getString("JOB_TITLE"));

System.out.println( "min salary = " +

rset.getInt("MINIMUM_SALARY"));

}

.

.

.


5.16.19 ResultSet.deleteRow() Behaviour Change


Prior to the V4.0 JDBC specification, the current row positioning after resultset-

based record deletion was not clearly defined in the JDBC specification.

In prior releases of Oracle JDBC for Rdb the row position would remain at the same

absolute position after the row has been removed. Thus, for example, if the deletion

row was the 10th

row, then after removing the record, the current row position would

still be the 10th

record within the resultset, effectively moving the remaining records

down one position.

This may cause unexpected behaviour while traversing the resultset. For example,

the following code will delete every second record in the resultset as after each

deletion the remaining rows would be shifted down one position:

.

.

.

Statement stmt = conn.createStatement(

ResultSet.TYPE_SCROLL_INSENSITIVE,ResultSet.CONCUR_UPDATABLE);

ResultSet rset = stmt.executeQuery ("SELECT * FROM employees");

while (rset.next ())

92

{

System.out.println("Select: " + rset.getString (1) );

rset.deleteRow();

}

.

.

.

To remove each row the row position would have to be reset each time back to the

prior row:

.

.

.

Statement stmt = conn.createStatement(

ResultSet.TYPE_SCROLL_INSENSITIVE,ResultSet.CONCUR_UPDATABLE);

ResultSet rset = stmt.executeQuery ("SELECT * FROM employees");

while (rset.next ())

{

System.out.println("Select: " + rset.getString (1) );

rset.deleteRow();

rset.previous();

}

.

.

.

In V4.0 of the JDBC specification, the row position after deletion using the

ResultSet.deleteRow() is now specified:

After the method deleteRow has been called, the cursor will be positioned before the next valid row.

If the deleted row is the last row, the cursor will be positioned after the last row.

Oracle JDBC for Rdb drivers have now been changed to comply with this behaviour.

The resetting of the row position using rset.previous() as shown in the example

above is no longer required.

5.16.20 Underlying Blob Handles not Released when using ResultSet.getBlob()


The ResultSet.getBlob() method may be used to return the contents of an Rdb

segmented string column to your application as a Blob object.

A problem in the underlying Rdb statements associated with the retrieval of

segmented strings prevented the handles from being released correctly.

93

This may lead to memory problems in the Oracle JDBC for Rdb server or the Java

application if many Blobs are being retrieved.

In addition as these statements are compiled and not released, operations such as

DROP TABLE may be blocked if attempted subsequent to and within the same

connection as Blob operations.

The following exception may be raised:

%RDB-E-NO_META_UPDATE, metadata update failed

-RDMS-F-ACTQUERY, there are queries compiled that reference relation

…


5.16.21 Sequence Values not Visible to ResultSet get Methods using Column Name


Bug 9659036

The retrieval of data from ResultSet columns using any of the ‘by name’ methods fail

when the column is one of the special sequence operations such as nextval or currval.

The following code will raise an exception when the second println statement is

called:

.

.

.

ResultSet rs = s.executeQuery(

"select reportseq.nextval from rdb$database" );

if (rs.next())

{

System.out.println(" by id = " + rs.getBigDecimal(1) );

System.out.println(" by name="+

rs.getBigDecimal("reportseq.nextval"));

}

.

.

.

The following exception will be raised:

94

Invalid column name

A similar problem may be seen when trying to retrieve column names using the

ResultSetMetaData.getColumnName() method:

.

.

.


"select reportseq.nextval from rdb$database" );

ResultSetMetaData rsMetaData = rs.getMetaData();

String columnName = rsMetaData.getColumnName(1);

while (rs.next ())

{

System.out.println("Select: " + rs.getBigDecimal(columnName));

}

rs.close();

.

.

.

The column name returned by the ResultSetMetaData method will be an empty

string and subsequent use of this value will fail to retrieve any column information.

As the first example shows, a work-around for the first problem is to use the column

index instead of the name in the get method.

An alternate work-around is to use a column alias in the query.

.

.

.


"select reportseq.nextval as NEXT_ID from rdb$database" );

if (rs.next())

{

System.out.println("by name = " +rs.getBigDecimal("NEXT_ID"));

}

.

.

.

There is no work-around for the ResultSetMetaData.getColumnName()problem.


95

5.16.22 Scaled Integer Problem


The Oracle JDBC for Rdb native driver may fail to return the values of scaled integer

columns correctly.

If the column size is less than 4 octets and has a scaling factor other than 0 the native

driver may return wrong values.

Columns with the following datatypes are affected:

● SMALLINT(n)

● TINYINT(n)

This problem only occurs when using the native driver; the thin driver will return the

values correctly.

SMALLINT and TINYINT columns not using scaling are not affected.


5.16.23 Server Matching Exception may Stop Poll Handling


If a server finds an exception during the handling of a server control function such as

a POLL request, the server will log the exception and then close down asynchronous

broadcast request handling.

This means that from that point onwards, operations such as ThinController POLLing

will no longer be acknowledged by the server.

Standard client and control function are not affected and the server will keep on

handling client requests correctly, however it will no longer respond to control

requests such as POLL.

A simple example of this behavior can be shown by issuing a server name matched

POLL request containing an invalid pattern string:

rdbthincontrol> poll #name:*srv*.

The server will log the following exception:

96

Dangling meta character '*' near index 0

*srv*.

^

The server will not respond to further POLL request until it has been restarted.

This behaviour has now been changed. The server will still log the exception but

unless the problem is a network IO problem or the server is in the process of shutting

down, the asynchronous broadcast request handling will still be enabled. Thus

operations such as POLL request handling will still be carried out by the server.

The only workaround is to restart the server again.


▲contents



5.17.1 Shutdown Thread

Starting with Oracle JDBC for Rdb release 7.3.0.0.0 (V7.3) the JDBC drivers will

create a shutdown thread when they are initially invoked. The purpose of this thread

is to use the shutdown-hook feature provided by OpenVMS that will execute the

thread at shutdown. The shutdown thread will ensure that any connection left open

by the application will be correctly disconnected prior to the application shutdown

proceeding, thus preventing application hangs at shutdown.

The application developer does not need to change any code for this shutdown feature

to be enabled, as long as the application is using Oracle JDBC for Rdb 7.3.0.0.0

driver libraries (or later versions) the shutdown hook will be in place.

See the section Shutdown Thread in the Oracle JDBC for Rdb User Guide for

details.

5.17.2 Driver.attach()

Release 7.3.0.0.0.

A new method has been added to the Oracle JDBC for Rdb drivers, that will allow

second and subsequent databases to be attached to an existing database connection.

97

See the Oracle JDBC for Rdb User Guide for details.

5.17.3 Returning List of Known Databases

Release 7.3.0.0.0.

The Oracle JDBC for Rdb Thin Driver will now allow client applications to obtain a

list of databases known to Oracle JDBC for Rdb servers.


5.17.4 Controller Enhancements

Release 7.3.0.0.0.

Several enhancements have been made to the Oracle JDBC for Rdb Thin Controller

application including:

● Obfuscate command

● Server matching patterns for the POLL command


5.17.5 RDB_EXT.JAR file

Release 7.3.0.0.0.

Oracle JDBC for Rdb release 7.3.0.00 installation now includes the RDB_EXT.JAR

file that contains classes and Java source for extensions to Oracle JDBC for Rdb.

The following extensions are include:

● Hibernate RdbDialect.java source

● Oracle UCP enabling classes

Details of these extensions follow.

5.17.5.1 Hibernate RdbDialect.java Source

Release 7.3.0.0.0.

Hibernate by Red Hat is a persistence engine that provides an alternative to standard

entity beans.

98

To allow Oracle JDBC for Rdb drivers to be used in conjunction with Hibernate the

RdbDialect.java source file is provided. You will find this source file in the top-level

folder of the RDB_EXT.jar.

See your Hibernate documentation for details on using third party JDBC drivers and

Dialects.

5.17.5.2 Oracle UCP Enabling Classes

Release 7.3.0.0.0.

UCP or, Universal Connection Pool, is a new database feature included in Oracle

database 11g 11.1.0.7, Oracle database 11.2.0.x and Oracle AS 11g R1. UCP works

with any Java based connections, e.g., JDBC, JCA, LDAP.

The RDB_EXT.JAR file installed with Oracle JDBC for Rdb, provides the necessary

classes to enable Oracle JDBC for Rdb drivers to be used in the connection pools

created by UCP.

The class:

oracle.rdb.jdbc.rdbExt.RdbDataSource

found in the RDB_EXT.JAR may be used as a Connection Factory class for UCP,

for example:

import oracle.ucp.jdbc.PoolDataSource;

import oracle.ucp.jdbc.PoolDataSourceFactory;

.

.

.

PoolDataSource pds = PoolDataSourceFactory.getPoolDataSource();

pds.setConnectionFactoryClassName(

"oracle.rdb.jdbc.rdbExt.RdbDataSource");

pds.setURL("jdbc:rdbThin://localhost:1701/my_dir:mf_personnel");

pds.setUser("jdbc_user");

pds.setPassword("jdbc_user");

pds.setInitialPoolSize(0);

Connection conn2 = null;

for (int i = 0; i< 5 ; i++ )

{

conn2 = pds.getConnection();

Statement sc = conn2.createStatement();

.

.

.

rs.close();

sc.close();

conn2.close();

}

.

.

.

99

For more details, see the UCP Users Guide at

http://www.oracle.com/technetwork/database/features/jdbc/index-091264.html

5.17.6 Performance Enhancements

Release 7.3.0.0.0.

Several enhancements have been made to the client/server connectivity and to

server/executor communications to improve performance including:

● Internal client/server message protocol changes

● Enhanced SQL Statement caching

● Results caching

● Server/executor synchronization

Details of these enhancements may be found in the following sub-sections.

5.17.6.1 Internal client/server message protocol changes

Release 7.3.0.0.0.

Network round-trip times for communication between client applications and JDBC

servers may constitute a large portion of wait time for applications. To help reduce

this wait time Oracle JDBC for Rdb has changed its client/server message protocol in

order to reduce the number of round-trips taken to:

● Attach to the database

● Compile and execute a SQL statement.

These enhancements are enabled automatically from V7.3 onwards.

5.17.6.2 Enhanced SQL Statement Caching

Release 7.3.0.0.0.

As well as caching of certain statement information on the client-side, reducing the

number of network IOs required for statement compilation, 7.3.0.0 allows the caching

of statement handles. See the sections SQL Statement Cache and Caching Statement

Handles in the Oracle JDBC for Rdb User Guide for more details.

5.17.6.3 Results Caching

Release 7.3.0.0.0.

http://www.oracle.com/technetwork/database/features/jdbc/index-091264.html

100

7.3.0.0 will allow limited caching of resultSets on the client side. This is particulary

useful in multi-tiered environments for statements that are repeatedly executed giving

the same results each time, especially in a pooled connection environment. See the

section Results Cache in the Oracle JDBC for Rdb User Guide for more details.

5.17.6.4 Server/Executor Synchronization

Release 7.3.0.0.0.

Starting with release 7.3.0.0.0, the method used to synchronize operations between a

Multi-process server and its sibling executors has been improved to reduce the

amount of CPU required and to speed-up operations.

▲contents



5.18.1 Server Startup Failure when using CFG File


A problem in parsing the configuration file during server startup may cause the server

to fail to startup correctly.

The following is an example of the error message seen:

$ java "-jar" "rdb$jdbc_home:rdbthinsrvpool.jar" –

"-cfg" rdbjdbc_cfg.cfg "-srv.mcGroupIP" "239.8.124.3" Configuration file problem at line 1 Content is not allowed in prolog.

This problem only occurs when using JDK 1.5-0 and above and using a Properties

file (*.cfg) to hold the server configuration properties.

A work-around for this problem is to use an XML-based configuration file instead.

101

5.18.2 AccessViolation on Disconnect when Inserting Blobs


A problem in determining the memory allocation for Blob variables may result in a

memory Access Violation during disconnect.

The problem only occurs if during the connection, data was inserted into Blob

variables and subsequently inserted into a segmented string column in the database.


5.18.3 PreparedStatement and Parameter Markers Known Problem now Resolved


In previous version of Oracle JDBC for Rdb the following known problem was noted:

Using PreparedStatement and Parameter Markers

During the creation of a prepared statement using the

Connection.PrepareStatement() method, the Oracle JDBC for Rdb

drivers call Oracle Rdb SQL to compile the SQL statement and describe its select

fields and parameter markers. At this time SQL builds internal message

representations of the parameter markers that may be passed to Oracle Rdb when

the prepared statement is executed.

The maximum size of character values that may be passed using each parameter

marker is fixed by SQL at this stage. This may cause inconsistent results when the

application attempts to use character string values that are longer than the

maximum size determined by SQL for that parameter.

If the input value is longer, the value will be truncated by SQL prior to being sent

to Oracle Rdb for processing.

This does not pose any problems if the query selection is equality, however, other

Boolean comparisons may cause unexpected results. For example, this query will

return the record:

.

.

. Statement stmt = conn.createStatement();

stmt.execute("create table tab (f1 char(3))");

102

stmt.execute("insert into tab values ('123')");

PreparedStatement ps;

ps = conn.prepareStatement( "select f1 from tab where f1 like ?");

ps.setString(1, "123");

ps.execute();

.

.

.

This query will not return the record:

.

.

.

ps.setString(1, "%123");

ps.execute();

.

.

.

The reason the above query fails is that SQL will set the maximum size of the

parameter text string to 3 characters (the size of field F1). The input value will be

truncated to %12 before being sent to Oracle Rdb and will not match the record.

In conjunction with changes made to Oracle Rdb 7.2.4.0.0 (V7.2-4), this problem has

now been resolved and queries similar to the ones shown above should now deliver

the correct results.

This fix will only work with Oracle Rdb versions 7.2.4.0.0 and higher.

5.18.4 Controller SHOW CLIENTS and MP Server Problem


When the command SHOW CLIENTS or SHOW ALL CLIENTS is issued within the

Controller, if the recipient server is a Multi-process server (MP Server), an executor

process may be started up by the server to execute the request.

A problem in the executor code prevents the executor from closing down correctly if

the maximum number of free executors has already been reached for the MP Server.

This new executor process will not be added to the free executor pool, but will not be

shutdown and thus will remain active on the system.

If the controller SHOW CLIENTS command is issued again, the number of executor

processes may increase until an Open VMS quota or process quota is exceeded.

Note

103

This problem was not completely fixed in the 7.3.0.0.0 release. This has

subsquently been rectified in the 7.3.0.1.0 release.

5.18.5 Possible Memory Leak when Updating Blob Columns


During the processing of update statements containing references to Blobs (list of

byte varying) columns, the JDBC drivers incorrectly compile the statement twice.

Only the second statement handle is released when the statement is closed which will

lead to Rdb statement handles still being active within the current connection.

These active handles use memory and resources within the drivers and the Rdb

system which may eventually lead to problems due to insufficient resources in long

running connections.

In addition these compiled queries may prevent metadata operations from occurring

on the affected tables within the current connection.

5.18.6 Access Violation with Trace and Network Dump


When the following trace flags are set together in TraceLevel, it is possible that an

access violation may occur if the size of the data being flushed to the network is very

large:

Bit Hexadecimal

Value

Decimal

Value

Traces

9 0x00000200 512 Network sends

30 0x40000000 1073741824 Full provides more details

on certain flags

For example, when inserting a Blob with more than 3MB of data the dump of the

network buffer may overflow memory and cause memory problems that will

terminate the server unexpectedly.

104

This problem only occurs with RdbThin connections using either a standard or a

Multi-process server.

5.18.7 Named Input Parameters not Working with CallableStatements


Using named parameters to set input parameter values of CallableStatements may fail

with the following exception:

oracle.rdb.jdbc.common.RdbException: Invalid column name :

The JDBC code fails to locate input parameters by name correctly.

An example of the type of code statement that may be affected:

.

.

.

stmt.executeUpdate("create module myproc " +

"LANGUAGE SQL " +

"procedure myproc(IN :val_in integer; OUT :val_out" +

"BEGIN " +

" SET :val_out = :val_in; " +

"END; END MODULE");

CallableStatement proc = conn.prepareCall(

"{ call myproc(?,?) }");

proc.setInt("val_in", 999);

.

.

.

Exception raised when executing the proc.setInt() method:

oracle.rdb.jdbc.common.RdbException: Invalid column name : val_in

Input parameters are marked as IN in the procedure definition.

Parameters marked as OUT or INOUT will be located correctly by the JDBC code.

A work-around for the problem is to specify the parameter index value instead of the

name.

105

▲contents



7.2.5.5.x.

None.

▲contents



7.2.5.5.x.

5.20.1 Long-running Query Holds up New Connections in MP Server


A problem in retrieving the Process Id of the current thread during the creation of a

new connection by the MP Server may cause the connection to hang.

This problem may be seen when an existing connection using the same MP Server is

concurrently preparing a very large SQL statement or is executing a query that is

taking substantial time within Rdb to fetch the very first record.

Any concurrent new connection requests will hang until the long-running query

returns the query compilation results or returns the first record back to the MP Server.

All other pre-established connections using the same MP Server will execute queries

normally and will not hang, only new connection requests are affected.


106

▲contents



7.2.5.4.x.

None.

▲contents



7.2.5.4.x.

5.22.1 Access Violation at Java_rdb_JNI_SetStrVal


A problem in a memory copy used within the JNI routine SetStrVal() may cause an

access violation similar to the following:

SYSTEM-F-ACCVIO, access violation, reason mask=00

.

.

.

%TRACE-F-TRACEBACK, symbolic stack dump follows

image module routine line rel PC abs PC

.

.

.

RDBJDBCSHR72 RDBJDBC Java_rdb_JNI_SetStrVal

.

.

.

This problem is infrequent as it depends on how memory is allocated during the

native code execution, and is usually only associated with JDBC code executing the

setString() method of a PreparedStatement or CallableStatement.


107

5.22.2 DCL Command Line Too Long


During the startup of a server either by a Pool Server or by using the Controller, DCL

command procedures are used to build the DCL command to invoke the server image.

It is possible that long directory names or deep directory structures associated with

the log file or the configuration files for the server may cause the size of the DCL

command created to exceed the length limits set for a command line by OpenVMS.

In this case the startup of the server will fail and the following exception may be

noted in the output of the server startup process:

%DCL-W-TKNOVF, command element is too long – shorten

The RDBJDBC_STARTSRV.COM has now been changed to reduce the number of

options that need to present on the command line by introducing another

configuration file that will be built each time the RDBJDBC_STATSRV.COM

command procedure is executed. This new command procedure will contain the

required configuration information necessary to correctly start the server.

5.22.3 Access Violation during DriverManager.getConnection() when Database Specification is Missing


If during a call to DriverManager.getConnection() the specified connection string

does not contain a database file specification a problem in the building of the

connection string to be sent to Rdb may force an unexpected termination of the

connecting server with the following exception:

An unexpected exception has been detected in native code outside the VM.

Unexpected Signal : EXCEPTION_ACCESS_VIOLATION (0xc0000005) occurred at

PC=0x18965EEB

Function=[Unknown.]Library= … \rdbjdbcshr.dll

NOTE: We are unable to locate the function name symbol for the error

just occurred. Please refer to release documentation for possible

reason and solutions.

Current Java thread:

at rdb.JNI.Connect(Native Method)

108

at

oracle.rdb.jdbc.common.AbstractNativeRdb.Connect(AbstractNativeRdb.java:301)

at

oracle.rdb.jdbc.srv.DBActionHandler.handleConnect(DBActionHandler.java:3844)

The actual exception raised depends on the platform and Java VM and may include:

#

# An unexpected error has been detected by HotSpot Virtual Machine:

#

# %SYSTEM-F-ACCVIO (0xc) at pc=84C1A4A0, pid=543163880, tid=63640192

and

SIGBUS 10* bus error

.

.

.

%SYSTEM-F-OPCCUS, opcode reserved to customer fault at PC=FFFFFFFF80B6EF14,

PS=0000001B

.

.

.

%SYSTEM-F-ACCVIO, access violation, reason mask=00, virtual address

=000000000000

0000, PC=FFFFFFFF80A708E8, PS=0000001B

and similar variants of the Access Violation messages.


5.22.4 Unaligned Memory Faults on IA64


Several unaligned memory faults have been located and fixed in the shared images

associated with Oracle JDBC for Rdb on IA64.

5.22.5 Read-Only Transactions not Enforced on Connection Switch


When the connection switch:

109

Transaction=readonly

is used, all transactions started by the Oracle JDBC for Rdb drivers should be forced

to READ-ONLY. However this is not correctly enforced if AUTOCOMMIT is

subsequently turned OFF and it is possible that READ-WRITE transactions may be

started on subsequent select statements.


5.22.6 Sockets not Correctly Closing on OpenVMS Clients causing Accumulation of Mailboxes


During the Close() of Connections using the Thin driver, sockets connecting

the client to the server will be closed and resources released. However a problem

with the release of data streams associated with these sockets prevents the sockets

from closing down completely.

The failure for the sockets to completely close down may cause problems on clients

running on OpenVMS. The mailboxes associated with TCP/IP sockets on OpenVMS

will not be closed and will accumulate for processes that do multiple connects and

disconnects using the JDBC Thin driver.

This problem only happens with clients running on OpenVMS and may depend on

the Java version and type of VM used.

The associated data streams are correctly closed down when the Connection class

is disposed, however unless explicitly disposed by the client application, the

Connection class object will continue to exist after the connection is closed and

until the JAVA garbage collection causes it to be disposed.

It is possible that the process may run out of TCP/IP ports or OpenVMS resources

due to this accumulation of mailbox channels, but this depends on how often the Java

garbage collection is run and whether or not the objects are collected.


5.22.7 Cast problem when Converting String to Date/Time


110

A problem in the JDBC driver code that converts String to the internal data format

required for storing to an Rdb date/time column may cause the following exception to

be raised.

java.lang.ClassCastException: java.util.Date

This problem occurs when a date/time value stored in its text form is used to set a

parameter in a PreparedStatement or CallableStatement that is associated with an Rdb

column or variable that has one of the following SQL datatypes:

● DATE ANSI

● DATE VMS

● TIME

● TIMESTAMP

For following code example will raise the exception described above.

.

.

.

// SQL : create table tabx (f1 date vms);

PreparedStatement ps = conn.prepareStatement

("insert into TABX values (?)");

String dtStr = "2003-11-07 12:34:56.780";

ps.setString(1,dtStr);

.

.

.

A workaround for this problem is to use a Java date/time conversion method to

convert from the String to a Java timestamp and then use the setTimestamp method

instead of the setString method.

String dtStr = "2003-11-07 12:34:56.780";

Timestamp dt = Timestamp.valueOf(dtStr);

ps.setTimestamp(1,dt);


▲contents

111



7.2.5.3.x.

None.

▲contents



7.2.5.3.x.

5.24.1 Interaction of DatabaseMetaData methods with Blobs may Crash the Thin Server


During the retrieval of RDB$DESCRIPTION data from Rdb System relations for the

inclusion into the resultSets returned by various methods within the

DatabaseMetaData class, the Oracle JDBC for Rdb Thin Server must create and

execute List Cursors.

A problem in the synchronization of access during List Cursor operations during

metadata retrieval meant that windows of opportunity exist where the thread currently

accessing the List Cursor may interfere with other concurrent threads accessing the

database from the same Thin Server.

This may result in variety of exceptions and/or bugchecks being raised. In some

cases the Thin Server may terminate unexpectedly with or without log or bugcheck

messages.

This problem only occurs within Thin Servers and only during DatabaseMetaData

method calls where description or comment data is being returned for the database

object, and only if there is another concurrent connection executing SQL statements.

For example, drilling-down Rdb database connection metadata within JDeveloper

using an Oracle JDBC for Rdb thin driver connection to a Thin Server may interfere

with concurrent clients on that same server.

112


5.24.2 BigDecimal scaling Incorrect when used with PreparedStatement SetObject() Methods


The scaling of BigDecimal objects may be incorrect when used as the source data

objects for PreparedStatement.SetObject() methods.

During the conversion of the BigDecimal to the underlying database datatype, scaling

information is lost which may result in the wrong values being applied.

A work-around for this problem is to use the

PreparedStatement.SetBigDecimal() methods instead.


5.24.3 Connection.nativeSQL() method Throws Null Pointer Exception


Calling the Connection.nativeSQL() method will result in a

NullPointerException exception being raised.


5.24.4 Calling Resultset.isLast() method May Change Transaction Behavior


This problem affects the Oracle JDBC for Rdb Native driver only. Applications using

the Oracle JDBC for Rdb Thin driver will not see this problem.

When the native driver is used, calling the Resultset.isLast() method will cause the

driver to fetch all the records of the resultSet to determine which is the last record.

During this processing the internal transaction status is set incorrectly. If a SQL

statement requiring a read-write transaction is executed subsequent to this but prior to

the resultSet's statement being closed, the update statement may fail with the


%RDB-E-READ_ONLY_TRANS, attempt to update during a read-only

113

transaction


▲contents



7.2.5.2.x.

5.25.1 DEC_KANJI and DEC_HANZI Support Enabled

Release 7.2.5.2.0.

Support was added to V7.1.3 Oracle JDBC for Rdb Drivers for accessing

DEC_KANJI and DEC_HANZI data from Oracle Rdb databases but until now had

not been adequately tested, so Oracle advised against using Oracle JDBC for Rdb

drivers to access DEC_KANJI and DEC_HANZI data from Oracle Rdb databases.

Testing of Oracle JDBC for Rdb drivers using these character sets in conjunction with

SHIFT_JIS on PC platforms has now been completed and the prior limitation of use

of these characters sets has now been removed.

▲contents



7.2.5.2.x.

5.26.1 ResultSet.getBigDecimal() not Working with System ResultSets


Calling the getBigDecimal() methods on a column in a resultset returned by

DatabaseMetaData methods may fail with the following exception:

java.lang.ArrayIndexOutOfBoundsException: 2

114


5.26.2 Setting TraceLevel fails when using Hexadecimal Notation


A problem in parsing hexadecimal values for trace level may cause an exception to be

raised when trace level values greater than 0x7FFFFFFF are used:



5.26.3 Delimited Identifier Problem in AS clause of Select Statement


A problem introduced in release 7.2.5.1.0 in parsing delimited identifiers used in the

AS clause of a select statement may cause an exception to be raised:

select last_name as "name 1", first_name from employees

SQLException: in <rdbjdbcsrv:prepare_stmt>

%SQL-F-RELNOTDEF, Table FIRST_NAME is not defined in database

or schema:42000

A work-around is to not use delimited identifiers in the AS clause:

select last_name as name1, first_name from employees


5.26.4 Configuration file problem in "DEFAULT" Server Definition


A problem in how properties were copied from the "DEFAULT" server definition

during the instantiation of server information may cause servers to have inappropriate

configuration settings.

115

This problem may only arise if the "DEFAULT" server definition in the server

configuration file contains any of the following properties:

ssl.default ssl.context

ssl.keyManagerFactory

ssl.keyStoreType

ssl.keyStore

ssl.keyStorePassword

ssl.trustStore

ssl.trustStorePassword

<allowPrivUser>

<allowUser>

<allowDatabase>

<allowPrivUser>

If any of the above properties are used in the "DEFAULT" server definition it is

possible that server configuration properties such as Allowed Databases, or Allowed

Users may be incorrectly propagated from the server they were specified for, to all

other servers described in the same configuration file.

A work around for this problem is to not use any of the above properties in the

"DEFAULT" server definition and instead place the property in each of the server

definitions that require it.

Note that the examples of configurations files used in the Oracle JDBC for Rdb

documentation may be susceptible to this problem as they do show the use of the

following property in the "DEFAULT" server definition:

ssl.default="true"

The problem has now been fixed.

5.26.5 Pool Server May choose Incompatible Pooled Server when User Restriction Enabled


If a server taking part in a pool of servers controlled by a Pool Server has restricted

access enabled and one or more AllowedUser entries are present for that server, the

Pool Server may choose the server as a possible candidate server for the connection

request, even if the connection username is not one of the specified AllowedUsers.

If this happens, the connection request will be redirected to that chosen server but the

connection attempt will be immediately terminated with the following exception:

SQLException: Access to server denied

116

Changes have now been made to ensure that information is passed correctly to the

Pool Server during the initial connection request to allow it to correctly determine if a

candidate pooled server will accept the user requesting the connection prior to

redirecting the connection request to that server.

5.26.6 Potential Problem when Dumping SQLDA in Trace


A problem in allocation of an internal buffer for the dumping of a SQLVAR data area

may cause unexpected and unusual problems during the execution of the server that

may result in the server terminating unexpectedly.

This problem may only be seen if the traceLevel DUMP SQLDA flag bit is set for the

connection or set server-wide and the data area of the SQLVAR being dump is

greater than 512 octets in length.

Bit Hexadecimal

Value

Decimal

Value

Traces

14 0x00004000 16384 Dump SQLDA information

A workaround for this problem is to not set the tracelevel DUMP SQLDA trace flag

bit.


5.26.7 Connection.getCatalog() Returns Wrong Value for Single Schema Databases


The Connection.getCatalog() method incorrectly returns a single quote delimited

string instead of a NULL object when connected to a single schema database. When

used in conjunction with a multi-schema database the correct catalog value is

returned.

This problem prevents JDeveloper 10.x from correctly displaying table and views

within the Database branch in the Connections Navigator.

A similar problem may occur when using the Connection.getSchema() method.

117

5.26.8 Potential Memory Leak with Views


A problem in the handling of statement preparation for SQL statements that contain

views where the view result tuples cannot have a dbkey associated with them, caused

a memory leak in the Oracle JDBC for Rdb drivers and servers.

The problem may manifest itself in a number of ways including access violations and

exceptions stating that shared memory has been used up.

The following example shows a typical view that may cause this problem:

create view view1 (c1 integer, c2 integer) as select c1,c2 from t1

union select c3,c4 from t2;

Queries on this view will execute correctly, however during the internal preparation

of each query, memory may be allocated that may stay allocated until the connection

is closed. In addition the underlying SQL statement may not be released correctly,

which in turn may prevent metadata updates from being carried out on the referred

tables.

▲contents



7.2.5.1.x.

5.27.1 SQLDA Dumping

Release 7.2.5.1.0.

Setting the tracelevel to 0x00004000 (Decimal 16384) will provide information about

the SQLDA information passed to and from SQL.


5.27.2 failSAFE IP with Pool Servers

Release 7.2.5.1.0.

Pool servers may be configured to ensure that redirected connection requests will still

correctly redirect during failSAFE IP fail over.

118


5.27.3 HandshakeTries and HandshakeWait on Multi-process Native Connections

Release 7.2.5.1.0.

The multi-process option on native connections allows the use of executor sub-

processes to carryout Rdb connections on behalf of your application using the

RdbNative driver. You now have the capability of specifying handshake options

during the initial communication handshake protocol used by the main and associated

sub-processes.


5.27.4 Server Access Security Enhancements

Release 7.2.5.1.0.

Servers may be configured to restrict the access of their served databases to a list of

allowed usernames. The server configuration allowUser has been added to the

server section of the configuration files restricting access to databases via that server

to only those users specified.

In addition a server password can be specified using the srv.password

configuration option which forces all users of that server to provide an addition

password before access via the server will be granted.


5.27.5 Restriction on using Multiple Blob fields in Join now Removed

Release 7.2.5.1.0.

In previous releases of Oracle JDBC for Rdb the following limitation was specified:

Blobs will only be returned correctly from a SQL join statements for the first table

mentioned in the join set. For example, given the SQL statement:

Select ta.blob, tb.blob from table1 ta, table2 tb

where ta.name = tb.name

ta.blob will be returned correctly as it is from the first table referenced in the join

set. Trying to access tb.blob may result in the following SQL error:

119

%SQL-F-BADPREPARE, Cannot use DESCRIBE or EXECUTE on a statement

that is not prepared

This restriction has now been lifted, the Oracle JDBC for Rdb drivers now handle

blob fields from multiple tables within a single join statement.

However due to the nature of the parsing carried out by the Oracle JDBC for Rdb

drivers it is required that all blob columns referenced from the second and subsequent

tables in the join must be qualified using correlation names as shown in the above

example of select.

Failure to use a correlation name in conjunction with the blob column name may

result in SQL parsing errors when data is retrieved from the blob field as the drivers

do not have enough information to determine the correct table to access the blob data

from.

SQL-F-FLDNOTCRS, Column <blob col> was not found in the tables in

current scope

This limitation also means that the use of "*" in the select clause for a join across two

or more tables that include blob fields may also cause a similar SQL error.

▲contents



7.2.5.1.x.

5.28.1 Incorrect Row Number Returned after ResultSet.getLast() call


A problem in the determination of the current row number when using Scrolling

ResultSets caused the ResultSet.getRow() method to return an incorrect row number

after absolute positioning of cursor after the end of stream.

The problem is only in the Oracle JDBC for Rdb Native Driver and does not show up

in when using the Oracle JDBC for Rdb Thin driver.

The problem may be seen only after a call has been made to ResultSet.afterLast()

method followed by a call to ResultSet.last().

120

The call to ResultSet.afterLast() incorrectly sets an internal record counter to one

greater than the actual count .

The following is an example of this problem.

.

.

.

Statement s2 = conn.createStatement(

ResultSet.TYPE_SCROLL_INSENSITIVE,

ResultSet.CONCUR_UPDATABLE);

ResultSet rs = s2.executeQuery("select * from employees");

rs.afterLast();

rs.last();

System.out.println("row number :" + rs.getRow());

System.out.println("employee_id :" + rs.getString(1));

rs.close();

s2.close();

.

.

.

May return the following information when used with the Employees table in the

PERSONNEL or MF_PERSONNEL databases provided as sample database in the

Oracle Rdb installation (the row number of the last record should be 100) :

row number :101

employee_id :00471

5.28.2 Pool Server Startup of Pooled Servers may fail When Persona is Used


A problem in the naming of the subordinate processes used to create a server process

during the automatic startup of servers by the Pool Server may cause the following

exception:

%RUN-F-CREPRC, process creation failed

-SYSTEM-F-DUPLNAM, duplicate name

The following related exception might also be seen during the attempted startup of

the pooled server process:

java.sql.SQLException: Unable to start process, status: 0x164 :

121

substatus -4

These problems may be seen only if PERSONA is used to change the server

authorization characteristics of the started servers.

5.28.3 Last Column in Select List may be Inaccessible in Some Queries


A problem in the handling of internal dbkey information may prevent the application

access to the last column in a select list. This problem only occurs if the select query

used cannot provide unique Dbkeys for the resultant tuples. Queries containing

derived tables or views from multiple tables may show this problem.

For example:

select c1.last_name from (select * from employees c where

c.employee_id='00170') c1

may fail to return the last_name correctly when the Resultset.GetString(String

columnName) method is used.

select c1.last_name, c1.first_name from (select * from employees c

where c.employee_id='00170') c1

may correctly return the last name but not the first name as the problem only affects

the last column in the outermost select list.

This problem was introduced in code changes made for release 7.1.3.0 of the Oracle

JDBC for Rdb drivers.

5.28.4 Abnormal Client Termination may Prevent Executor Re-use


If a client application using a MP Server terminates abnormally or the client socket is

lost, the associated database connection will be disconnected by the server however

due to an internal problem, the executor process associated with the terminated client

may remain present on the system in LEF state.

The handling of abnormal termination did not correctly terminate the executor

process, nor did it place the free executor back in the free list for re-use. This results

122

in orphaned executor processes that will remain on the system in LEF waiting state

but will never be re-used.

If the client abnormal terminations occur frequently, the number of inactive executor

processes will grow and may eventually cause system resource problems and

excessive swapping.

5.28.5 Decimal Column Problem with Native Driver


A problem in the nativeRdb driver caused Decimal columns to be returned

incorrectly. This problem only affects applications using the rdbNative driver.

The Decimal datatype may be used by SQL when scaled integers are returned after

manipulation by an internal function or aggregate operation, for example the

following query may return incorrect values when executed through the rdbNative

driver.

select distinct salary_amount from salary_history

Application using the rdbThin driver should not encounter this problem.

5.28.6 'EFN xx is not available' Message on Executor Startup

Changed in Instance Build 20070302 release 7.2.5.1.0.

In earlier versions of Oracle JDBC for Rdb, if the MP Server found that the common

event flag number it was using to start the handshake process with a newly created

executor, was not available after trying to obtain it for reasonable length of time, the

server would abort the client connection attempt with the following exception:

'EFN xx is not available'

In such a case the event flag used by the server was probably left set by a previous

attempt by the server to create an executor process but failed during the process

startup due to resource problems.

The server code has now been changed to correctly clean up its event flags usage if an

executor process fails to start up correctly. In addition the server now does not abort

when it finds the event flags already in use, but now assumes that as the amount of

time the event flag is unavailable is much longer than the amount of time it might

123

take the executor process to be created and executor image run-up, that the flag may

be reset and the current executor start-up may proceed.

5.28.7 Extraneous log message during Auto-restart check by Pool Server

Changed in Instance Build 20070306 release 7.2.5.1.0.

A problem in how the Pool Server checked the availability of its pooled servers when

carrying out AutoRestart checking meant that an extraneous CLIENT LOST message

would be logged by the pooled server every time it was checked. The following

message would be logged:

srv.DBActionHandler <idle> Connection to Client lost


5.28.8 Logfile not Correctly set for Servers Started Using the Controller


The logfile used by the server to record trace message and other output may be set in

the server specification section of the configuration file used in conjunction with the

servers and the Controller.

A problem in how the logfile information was passed to the newly started server

process prevented the correct logfile specification to be used by the server when the

server was started using the Controller Start Server command.


▲contents


5.29.1 Persona

Release 7.2.5.0.0.

When a Thin or Pool server starts up, it automatically inherits the rights identifiers,

quotas, and authorization attributes of the process under which it was started. You

may now override this default behavior by specifying a persona to use on the startup

of the server. This persona will then be used by both the server and the underlying

124

OpenVMS operating system to determine the rights and authorities of the server

process and any executor processes that the server may start up.

This feature was introduced in release 7.1.4.0.0, but was omitted from the release

notes.



7.2.5.0.x.

5.30.1 Incorrect SQLSRV_JDBC_SERVER_STARTUP72 Installed with V7.2-41 Oracle JDBC for Rdb Kit


An incorrect version of the SQLSRV_JDBC_SERVER_STARTUP72.COM file was

inadvertently placed in the installation kit for Oracle JDBC for Rdb 7.2.4.1.0.

This version of the file does not set up the RDB$JDBC_SQSNAM_* logical name

properly and may cause problems when you try to use this file with SQL/Service Thin

server startup.

The following line of DCL command in this file is incorrect:

$ nam :='f$logical("RDB$JDBC_SQSNAM_''port'")

It should read:

$ nam = f$logical("RDB$JDBC_SQSNAM_''port'")


5.30.2 Multi-process Server May Show Continuous DIO Activity Even When Idle


A problem with the way error and output channels are assigned during the creation of

the executor subprocess by a detached Multi-process server may cause the server

process to continually issue direct I/Os to the associated mailboxes. This can be seen

125

as a continuous rise in the "Direct I/O" count for that process even when the server is

idle.

Although this does not interfere with the correct functionality of the server, it could

incorrectly show up as activity on a quiet server.

A workaround is to start up the Multi-process server directly in a login session rather

than detached.


5.30.3 Client idleTimeout Does Not Work for Prestarted and Reused Executors


The amount of time that a client connection may be idle can be limited by using the

cli.idleTimeout parameter for the Thin server.

However, the client idle timeout value set for the server will be ignored when a Multi-

process server is used with prestarted executors.

If the client gets a prestarted executor on connection, the client idle timeout for the

server does not get properly transferred to the client context and no timeout will be

issued.

Additionally even if an executor was not prestarted, if it is reused then a similar

problem will occur and the inactivity timer will not be set.

The client idle timeout set for a server is now correctly observed by prestarted and re-

used executors.

5.30.4 Syntax Error in Query Generated for DatabaseMetaData.getTables


The JDBC DatabaseMetaData.getTables() method allows the caller to obtain

information about the tables and views found in a connected database. When you call

this method, you can supply a list of table types to search for.

Currently the Oracle JDBC for Rdb drivers recognize the following types of tables for

this method:

● TABLE

● VIEW

126

● SYSTEM

● SYSTEM TABLE

● SYSTEM VIEW

● LOCAL TEMPORARY

● LOCAL TEMPORARY TABLE

● GLOBAL TEMPORARY

● GLOBAL TEMPORARY TABLE

● INFORMATION

● INFORMATION TABLE

The drivers should ignore any table type not in the above list.

However, due to a problem in the driver code, if the list of table types starts with a

type that is not recognized by the driver, a SQL syntax exception will be generated.

For example, the following example will result in a SQL syntax error:

String types[] = {DERIVED, "TABLE", "VIEW", "GLOBAL TEMPORARY"};

ResultSet rs = dbmd.getTables("","", "%" , types);

One possible workaround for this problem is to re-order the types so that the first type

specified is one from the list of recognized table types, for example:

String types[] = ("TABLE", DERIVED, "VIEW", "GLOBAL TEMPORARY"};

ResultSet rs = dbmd.getTables("","", "%" , types);

This example does not generate a SQL error.


5.30.5 Show Clients in Controller may Crash Connected Thin Server


A change in handshake protocol in Oracle JDBC for Rdb 7.1.4.1.0 (V7.1-41) drivers

introduced a problem in how thin servers respond to requests for client information.

Issuing a SHOW CLIENT command in the Oracle JDBC for Rdb Controller

command line may cause the connected thin server to access violate and consequently

terminate the server process.


▲contents

127


This section contains new features and technical changes for Oracle JDBC for Rdb

release 7.2.4.1.x.

5.31.1 Client and Server Timeout Feature

Release 7.2.4.1.0.

You can now specify the amount of time a server or a client connection may remain

inactive before the connection will be terminated or the server closed down.


5.31.2 Executor Name Prefix

Release 7.2.4.1.0.

You can now specify the name prefix for executors started up by the Multi-process

server. This can help in identifying executor processes on your system.


▲contents



7.2.4.1.x.

5.32.1 Release Notes Specify Incorrect Installation Directory for RDBJDBCCFG.XML


The release notes for Oracle JDBC for Rdb release 7.1.2.0.0 incorrectly specified that

the RDBJDBCCFG.XML file would be copied to SYS$COMMON:[RDB$JDBC]

directory. The RDBJDBCCFG.XML is actually coped to two directories during

product installation:

● The product installation directory found under the main JDBC directory, for

example,

SYS$COMMON:[RDB$JDBC.0701-4V0614]

128

● The SYS$COMMON:[RDB$JDBC.COM] directory

In addition, the installation procedure incorrectly replaced the RDBJDBCCFG.XML

file in the SYS$COMMON:[RDB$JDBC.COM] directory, overwriting any already

existing file of the same name.

The release notes have been fixed, and the installation procedure will only copy the

RDBJDBCCFG.XML file to the SYS$COMMON:[RDB$JDBC.COM] directory if

the file does not already exist in that directory.

5.32.2 Persona Not Handled Correctly by the Multi-process and Pool Servers


When the Persona feature is used in conjunction with a Multi-process or Pool server,

a problem in the way either the executor processes or the pooled server processes are

created prevented the correct Persona identification from being passed to the created

processes. This problem may result in the following error being raised:

java.io.IOException: Child creation error: not owner

Due to a restriction in the use of the Java System.exec() method that was used by the

JDBC servers to start executor sub-processes and pooled servers, the security

information and Persona details were not copied across to the newly created process.

The JDBC servers now use the OpenVMS system service CREPRC to start processes.

CREPRC correctly transfers the security information to the new process.

5.32.3 Multi-process Server / Executor Handshake Timeout May Be Too Short on Heavily Loaded Systems


When a Multi-process server talks to an executor it uses a handshake protocol to

check that the executor is still alive and accepting direction. By default, if the

executor has not responded to the server's synchronization request within five seconds

it will raise the following exception and terminate the connection :

Lost connection to executor

129

This synchronization handshake is done after the executor has replied to the server

that it has completed the task requested and is waiting for the next operation to carry

out. This synchronization failure will not be raised while the executor is busy within

the database and thus is unaffected by such things as database locks or the duration

required to compile or execute queries. It will only occur when the executor is known

to be waiting for the next action to carry out.

In heavily loaded systems, especially on single-cpu systems, it is possible that the

executor process may not be scheduled for execution within the window of this

synchronization handshake and the exception may be raised.

In order to carry out this synchronization, in previous version of the drivers the server

polls the executor up to 500 times with a 10 millisecond delay between each poll

request. If no response is found after 500 tries, the server raises the above exception.

This version of the Oracle JDBC for Rdb drivers now allows you to specify, at the

server level, the maximum number of poll tries and the delay between each try. If you

know that the system on which the server is executing could possibly have extended

process scheduling delays, you can ensure that the server will not time out on the

synchronization handshake. Two new switches have been added to the server

definition and startup.

● Srv.MPmaxTries---Use to specify the maximum number of poll tries

● Srv.MPtryWait---Use to specify the delay between each try

See the Oracle JDBC for Rdb User Guide for more information.

5.32.4 Problems with srv.idleTimeout and srv.bindTimout Configuration Variables and Their Use with SSL servers


The Oracle JDBC for Rdb User Guide for release 7.1.3.0.0 (V7.1-3) incorrectly

referred to the srv.idleTimeout as affecting the inactivity timeout for a connection.

This switch actually refers to the timeout period for server inactivity. In addition, this

feature was not fully functional in previous versions.

The srv.bindTimeout configuration variable was meant to limit the time the server

will wait for an acknowledgement from the client that the database attach should

proceed. The default value is 0, which means that the server will wait indefinitely.

This timeout is useful when dealing with SSL communication, as the server uses it to

limit the time it will wait for the client to send down an attach request after a new

socket connection has been requested. If the client fails to use an SSL secure socket

when trying to communicate with a server that has SSL enabled, the client thread

130

within the server will hang as the connection cannot complete. The srv.bindTimeout

value specifies how long this wait should be before giving up.

Unfortunately the default for the value was incorrectly set to 1 second, and the

srv.bindTimeout server attribute was ignored in the XML configuration file. This

meant that on CPU-bound systems it was possible that the initial SSL negotiation

could take longer than one second and thus cause a TIMOUT failure on the new

connection request.

These problems have now been fixed. See the Oracle JDBC for Rdb User Guide for

more information.

5.32.5 IA64 Problem Causes Array Out of Bounds Exception When Handling String Indexing


A problem in the way Java on IA64 carries out string index operations in association

with static final string constants may infrequently cause the following type of

exception to be raised:

Caused by: java.lang.ArrayIndexOutOfBoundsException: 1054649176 at

java.lang.String.indexOf(String.java:1266) at



oracle.rdb.jdbc.common.Statement.getTableName(Statement.java:3148)

In all cases, the index value shown after the exception name is very large, in the same

order of magnitude as seen above.

The getTableName method has now been changed to a mechanism other than indexOf

to carry out its operation. This problem should no longer be seen.

5.32.6 Comments within SQL Text Not Handled Correctly


Executing or preparing a statement that has SQL text containing leading or embedded

comments may cause errors during parsing of the statement.

Some third-party products may use comments such as /* comment */ in the text they

send down to the JDBC drivers for compilation. Although handled correctly by

Oracle Rdb, comments of this style caused a problem in the determination of

statement types during the preliminary parsing of the statement by the JDBC driver.

131

For example the following SQL text:

stmt.Execute(/* This is a comment */ select * from jobs);

would cause an SQLException:

SQLException: in <rdbjdbcsrv:execute_immediate> %SQL-F-EXESELSTA,

Attempted to EXECUTE a SELECT statement:RR000

The JDBC driver could not correctly determine the type of statement and used the

wrong underlying SQL operation to attempt to execute it.

The drivers now extract out comments prior to determining the statement type and

sending the native SQL down to Oracle Rdb. The drivers will now correctly parse out

C and SQL type comments, for example:

/* comment */

! this comment will be terminated at the next line break

-- this comment will be terminated at the next line break

// this comment will be terminated at the next line break

5.32.7 Prepared Statements May Cause a Memory Leak with Multi-process Servers


During the preparation of PreparedStatements, the Multi-process server has to

allocate memory from the servers' global shared memory pool that will hold some

information about columns and parameter markers in the statement that is being

prepared.

Due to a coding problem, some of this memory was incorrectly allocated each time

the prepared statement was executed, instead of only once at statement compilation

time. This wrongly allocated memory was never freed after use. Executing the same

prepared statement multiple times will slowly diminish the shared memory available

to the server, eventually causing a problem when the shared memory allocation is all

used up.


▲contents

132



5.33.1 Maximum Size of Single Data Row Increased to 65,272 Octets


During copying rows of data from Oracle Rdb, the Oracle JDBC for Rdb drivers

incorrectly limited the number of octets copied to 36863 octets. This can cause

problems when there are more than 36863 octets in the row.

The following exception is a symptom of this data row truncation:

Statement creation failed: java.sql.SQLException: Connection lost :

java.lang.NegativeArraySizeException @rdb.Client.fillCache

The maximum size of a data row supported by the drivers has now been increased to

65,272 octets in keeping with the maximum row size supported by Oracle Rdb.

5.33.2 Another Connection Overlap Window Found with Pool Servers


Another potential overlap of connections between the connection made by the Oracle

JDBC for Rdb Pool server and its pooled servers has been found which may cause the

incorrect rejection of a client connection even when a free connection slot is

available.

This is similar to the problem referred to in the Oracle JDBC for Rdb 7.1.3.3.0 release

notes as Spurious Maximum Number of Clients Exceeded Exception.

The handshake protocol during server check by the Pool server has now been

changed to prevent this overlap of connections.

5.33.3 SSL Server Information Not Correctly Set from XML-Formatted Configuration File


A problem in the parsing of XML configuration file data prevented the correct port

and node information from being assigned to named servers of the type

"RdbThinSrvSSL", "RdbThinSrvMPSSL" and "RdbThinSrvPoolSSL.

133

Any URL specification provided for the individual server would be ignored, and the

default port and node used instead.


▲contents

oracle® jdbc for rdbdownload.oracle.com/otndocs/products/rdb/pdf/rdbjdbc_7341_rn.pdf · 4.2 using...

Documents